This study used four observation materials. The substrate allyloxybenzene (1a) was obtained from commercial sources (WAKO Chemical Ltd.). The substrates 1-allyloxynaphthalene (2a), 1-crotyloxynaphthalene (3a) and 1-cinnamyloxynaph- thalene (4a) were synthesized based on a previous report [9] . The Claisen rearrangement was performed by mixing the substrate, e.g., 1a, 2a, 3a, 4a (0.5 mmol), FeCl₃ catalyst (0.15 molar ratio) and the solvent including 1,2-dichlo- roethane and decalin (7 cm³) in a test tube. The usage of 1,2-dichloroethane re- presenting as a low polar solvent and decalin as a nonpolar solvent are to examine the solvent effect on this reaction. The mixtures were heated at the low temperature of 80˚C with stirring in a microwave reactor or in an oil bath. The microwave experiment was carried out using a single mode microwave apparatus (Green Motif II) operated at 300 W and 500 W. The reaction was conducted under an argon gas atmosphere. The reaction mixtures were sampled at 2, 5, 20, and 40 minutes. After 60 minutes, the reaction mixtures were cooled to room temperature. The mixtures were then filtered and washed using ion-exchange water. When decalin was used as the solvent, the mixtures were extracted using methanol.

The reaction mixtures (0.5 mmol substrate, 0.15 molar ratio FeCl₃ catalyst and 7 cm³ of 1,2-dichloroethane) were reacted under the microwave irradiation conditions. At a specific time, a 0.1 cm³ sample was taken out by a syringe with a long needle. The reactions were performed at several temperatures of 50˚C, 60˚C, 70˚C and 80˚C.

The ratio of the products was identified by High Performance Liquid Chromatography (HPLC) analysis using a Shimadzu LC-20A system equipped with a UV detector. An ODS column was used with methanol: water (9:1, v/v) as the eluent. The structure of the products was confirmed by Nuclear Magnetic Resonance (NMR) and Gas Chromatography Mass Spectrometer (GC-MS) analysis. The NMR spectra were recorded by an FT-NMR ECS 400 (JEOL) at 400 MHz (¹H) and room temperature in CDCl₃. The molecular weight and mass fragmentation of the products were determined using a GC-MS QP5050A from Shimadzu. The ¹H-NMR relaxation time was measured by the inversion-recovery method to investigate the molecular motion of the proton.

Initially, the Claisen rearrangement of the allyloxybenzene (1a) was investigated in the absence of the FeCl₃ catalyst. As the result, the reaction could not take place without the FeCl₃ catalyst at 80˚C under microwave irradiation and no product was observed. In the presence of the FeCl₃ catalyst, the reaction proceeded. Based on previous reports, the utilization of a metal catalyst induced a sequential reaction [10] . In this reaction using FeCl₃, both a rearrangement (1b) and cyclization product (1c) were obtained. When 2-allylphenol (1b) was heated in the presence of the FeCl₃ catalyst, 2-methylcoumaran (1c) was formed as the sole product.

The reaction mechanism was suggested through rearrangement of 1a affords 1b, followed by cyclization of 1b leading to 1c (Scheme 1). It was postulated that FeCl₃ catalyzed the rearrangement and cyclization reaction through coordination to the oxygen atom at the transition state. FeCl₃ facilitated the reaction by causing an inductive effect that would effectively polarize the transition state. This interaction would decrease the activation energy so that the reaction could proceed at low temperature.

Scheme 1. The rearrangement and cyclization reaction pathway.

Before investigating the more complex structures, the kinetic behaviors of the reactions were studied. The decreasing amounts of 1a and increasing amounts of 1c were measured as a function of time at four different temperatures between 50˚C and 80˚C in 1,2-dichloroethane. The plots of the natural logarithms of the concentrations of 1a and 1c versus time at each temperature are shown in Figure 1 and Figure 2.

From Figure 1 and Figure 2, it is clear that the linear correlative coefficients are higher than 0.98, which indicates that ln[A] is proportional to the reaction time at different temperatures. It was showed that the reaction followed the first-order reaction. The measured rate constants were obtained from the slope values of the straight line and summarized in Table 1. Using the measured rate constants, the activation energy (Ea) could be obtained. The parameters of the Arrhenius equation are obtained from the relationship of the rate constant with the reaction temperature. The Arrhenius plot for the Claisen rearrangement of 1a and the cyclization reaction of 1b are shown in Figure 3.

The Ea values of the rearrangement and cyclization reactions calculated from the slope were 30.22 kJ/mol and 85.92 kJ/mol, respectively. The Ea value of the cyclization reaction was higher than that of the Claisen rearrangement reaction. The proposed energy profile showed that the cyclization reaction of 1b required a higher energy to proceed than the Claisen rearrangement reaction of 1a (Figure 4). Thus, it indicated that the rearrangement reaction is kinetically controlled and the cyclization reaction is thermodynamically controlled.

It was then necessary to determine if the reaction was proceeding through an intermolecular mechanism in which the substrate became fragmented into separate nuchleophilic and electrophilic species or through an intramolecular mechanism. Previously, our laboratory reported that naphthalene derivatives have an intermolecular rearrangement character [9] . Thus, we expanded the scope of the reaction with 1-allyloxynaphthalene (2a). The rearrangements of the naphthalene derivatives bearing a crotyloxy group (3a) and a cinnamyl group (4a) at the 1 position were also discussed. The products are shown in Table 2.

As the results, the main product of the reaction of 3a and 4a were the intramolecular rearranged one even in the presence of FeCl₃. The intermolecular rearranged product was not observed in the reaction mixtures. This clearly showed that the reaction occurs through the intramolecular mechanism and not the intermolecular mechanism. The authors also reported the use of a crotyloxy group in the aromatic Claisen rearrangement, which is not a well-known starting material for the aromatic Claisen rearrangement unlike the cinnamyloxy group. The product yields of 3a were higher than those of 1a and 2a, however, the dehydrogenation product (3d) was formed. The methyl substituent acts as an electron donating group, thus enhancing the electron density of the carbon atom where the formation of new bond will take place. Therefore, the yield increased for the reaction of 3a. In this case, the additional methyl substituent on the allyl moiety also leads to the dehydrogenation reaction of the coumaran derivatives.

^aSolvent: 1,2-dichloroethane, ^bMethod: microwave 300 W, ^cReaction time: 60 minutes, ^dThe yields were determined by HPLC.

Interestingly, the reaction of 4a only produced 4b as the main product in an excellent yield. The cyclization product was not formed in the reaction of 4a. On the other hand, our previous studies showed that the para-rearranged product is accelerated for cinnamyloxyarene derivatives. Because the stable conformation for the cinnamyloxyarene derivatives, in which the C-C double bond is near the naphthalene ring and the phenyl group is situated away from the naphthalene ring, is easily converted in the transition state to the para-rearranged product [9] [11] .

In this study, we suggested that the phenyl ring blocked the binding site between the oxygen atom and the double bond during the reaction. As shown in Figure 5, FeCl₃ has the possibility to interact with the phenyl ring besides coordinating with the oxygen atom. Since the aromatic ring is an electron-rich π system, Fe³⁺ as a metal cation is able to interact with the aromatic ring. Thus, the cyclization reaction did not occur and the para-rearranged product was not formed due to the opposite conformation of the C-C double bond. Hence, the ortho-rearranged product would remain as the sole product.

The spin-lattice relaxation time (T₁) can be used to determine the conformational preference of the molecules [12] . Therefore, the ¹H-NMR relaxation time study was conducted to investigate the molecular motion of the proton. The T₁ of the proton at the C₃ position was measured, since the C₃ proton should be influenced by the motion of the substituents in the allyl group. The results are shown in Figure 6.

Based on these results, the T₁ of the C₃ proton of 4b was determined to be 9.66 s. This T₁ value is higher compared to those of the other two substrates. The high T₁ suggested that the C₃ proton has a high mobility and the phenyl group lies away from the C₃ proton. In the case of 3b, the rotation around the C₂-Cγ bond axis hinders the motion of the C₃ proton, thus the relaxation time would be short.

In this study, the comparison between the microwave irradiation and oil bath heating for the Claisen rearrangement reaction of allyloxyarenes is discussed. The results are summarized below in Figure 7 and Figure 8.

As can be seen in Figure 7, the accelerating effect during microwave irradiation was observed for substrates 2a and 4a in decalin. Especially for product 4b,

the yield from the microwave 500 W is double that for the oil bath heating. This would be due to the fact that decalin is a nonpolar solvent which does not absorb the microwaves, so the accelerating effect occurs on the polar substrates. However, the microwave accelerating effect was not observed in the Claisen rearrangement of 3a. The additional methyl substituent which attribute to the high reactivity might shade the microwave effect. In the case of 1a reaction, an increase in the microwave power was necessary to increase the product yields. Since decalin has a low dielectric constant, thus a higher energy to promote the reaction during the microwave irradiation was required.

1,2-Dichloroethane has a low polarity compared to decalin. Based on a previous report, the Claisen rearrangement is accelerated by a more polar solvent, because the transition state, in which the oxygen atom has a higher electron density, is stabilized by the coordination of the solvent [13] . As a consequence, the reaction yields would be higher than in decalin. However, such a solvent more efficiently absorbs the microwaves, so increasing the microwave power did not significantly affect the yields in the 1,2-dichroloethane (Figure 8). Interestingly, in the Claisen rearrangement reaction of 3a, oil bath heating afforded higher yield in the dehydrogenation product (3d) instead lower yields of the Claisen rearranged product (3b).

The temperature change in the reaction system under microwave irradiation was found. The substrates 1a and 2a were dissolved in the solvent and irradiated by the microwave. Then temperature versus microwave irradiation time was measured at the same conditions for the reaction system. The results are shown in Figure 9 and Figure 10. In the presence of the FeCl₃ catalyst, the temperature increased faster than in the absence of FeCl₃ catalyst; the substrate alone exists in decalin or 1,2-dichloroethane. These results indicated that the microwave energy is absorbed by an active species in the reaction mixture. This suspected active species is the adduct between the substrate and FeCl₃ catalyst. FeCl₃ as a Lewis acid would forms a complex with the aromatic ring as shown in Figure 11. The

complex species was expected to efficiently absorb the microwaves and affected the temperature. The high temperature in the micro region around the species would influence the acceleration effect during the microwave heating reaction.

Based on a ¹H-NMR relaxation time study, the expected formation of the ferric-arenes complex was elucidated. Generally, shortening of the proton relaxation time indicates a decreasing of the mobility of the proton [14] . As shown in Table 3, in the presence of the FeCl₃ catalyst, the relaxation times of 1a and 2a became shorter than that for no FeCl₃ catalyst. In this case, the aromatic rings of 1a and 2a interacted with the Fe³⁺ and formed a complex. The species have a larger volume, thus they more slowly rotate. Consequently, the mobility of the proton in the aromatic ring would be reduced. In non-aromatic substrates, e.g., cyclohexane and decalin, such a decrease in T₁ did not occur. This means that FeCl₃ does not interact with the cyclic ring of cyclohexane and decalin.