The Reaction of Safrole Derivatives with Aluminum Chloride : Improved Procedures for the Preparation of Catechols or their mono-O-Methylated Derivatives and a Mechanistic Interpretation

Um procedimento otimizado para a preparação de catecóis através da reação de quebra da ponte metilenodioxílica de derivados do safrol com cloreto de alumínio é descrito. Em substratos com grupos substituintes atraentes de elétrons (carboxaldeído e nitro) observou-se a formação regiosseletiva de éteres clorometílicos facilmente isoláveis. A partir desses intermediários foram sintetizados os fenóis mono O-metilados (3-hidroxi-4-metoxibenzaldeído, 2-bromo-4-metoxi-5hidroxibenzaldeído, 2-nitro-4-metoxi-5-hidroxibenzaldeído, e 2-metoxi-4-(2-oxoprop-1-il)-5nitrofenol). Com base nesses dados experimentais e em cálculos semi-empíricos (MNDO) de orbitais moleculares foi proposta uma racionalização mecanística para explicar as regiosseletividades observadas.


Introduction
The 3,4-dihydroxyphenyl group 1 and its O-substituted analogues 2 and 3 (Figure 1) are commonly found in numerous naturally occurring products 1 and pharmaceutical compounds 2 .For the preparation of these compounds, the availability of starting materials with appropriate patterns of substitution is commonly required.Safrole (4a, Figure 1), an abundant phenylpropanoid bearing a methylenedioxy ring, and its easily available derivatives 4b and 4c have been used for this aim 1,2 .As the cleavage of the methylenedioxy ring is a necessary step to be carried out sometime during the processes of conversion of 4 into products having the structural pattern present in 1, 2 and 3, several methods with this purpose have been developed 3 .Most of them have been used only for preparations of free catechols (1-type products), as exemplified by almost all of the methods mediated by electrophilic reagents 3 .On the other hand, when the aromatic ring in derivatives of 4 contains electron-withdrawing groups (as in 4b, c), the methylenedioxy ring can be regioselectively cleaved by nucleophilic reagents (RO -or ArO -) leading either to palkoxy products (2-type; resulting form ipso attack of RO) or to m-aryloxymethoxy derivatives (3-type, resulting form attack of ArO -at the methylene carbon) 3,4 .Similar results have also been reported for some electrophilic reagents, as ether free Grignard reagents 3,5 , but low regioselectivity has resulted 5 .Furthermore, AlCl 3 reacts with piperonal derivatives 4c and 4d, in CH 2 Cl 2 at room temperature, followed by treatment with refluxing aqueous HCl/THF/ NaI or KI, leading to the corresponding catechols 6 .
In this paper we describe our results on the reaction of safrole derivatives 4a-h (Schemes 1 and 2) with AlCl 3 : (i) an improved procedure to obtain the corresponding catechols 11b, e in a one-pot procedure and mild conditions (Scheme 1), (ii) the use of the chloro ether intermediates 5b-d and 6e in the regioselective synthesis of mono-O-methylated phenols 9b-d and 10e (Scheme 2), and (iii) a theoretical study of the reaction course by semi-empirical (MNDO) 7 molecular orbital calculations.Based on these new experimental and theoretical data, we were able to propose a mechanistic rationalization for the observed results.

Results and Discussion
Compounds 4c-h were prepared from 4a and 4b as previously described 8 .Substrates (4a-h) were reacted with AlCl 3 in CH 2 Cl 2 at room temperature and the reaction mixtures were quenched in two different ways: (a) with cold water, followed by overnight stirring at room temperature (Scheme 1) and (b) with glacial AcOH (Scheme 2).When H 2 O was used to quench the reaction, the resulting biphase mixture was stirred overnight at room temperature and the catechols 11b, e were isolated as pure compounds (> 95% by 1 H NMR) simply by extraction of the aqueous layer with EtOAc (Scheme 1).This constitutes a one-pot procedure for the preparation of these catechols from 4b, e under mild conditions 6 and high degree of purity, precluding the isolation of the chloro methyl ether intermediates 6a,b .
When glacial AcOH was used (Scheme 2), the relatively unstable chloro methyl ether intermediates (5 and/or 6) were isolated.In the case of 4f and 4g an almost equimolecular mixture of 5f and 6f was obtained 9a .We were able to observe that epoxide 4g is converted into ketone 4f faster than the cleavage of the methylenedioxy ring 9b .On the other hand, only one regioisomer was observed when 4b-e were used as substrates.Nevertheless, the use of this procedure with 4a and 4h has led to an intractable mixture of products.In order to establish the structure of the regioisomer, the isolated intermediates were methylated under neutral conditions (CH 2 N 2 , Et 2 O, cat.neutral Al 2 O 3 ) and the products, 7 or 8, were smoothly hydrolyzed (AgNO 3 , THF, H 2 O) to the corresponding o-methoxyphenols (9 or 10; Scheme 2).Based on spectral data and melting points, we were able to unambiguously assign the structures 9b, 9c, and 9d for the o-methoxyphenols formed from 4b, 4c, and 4d, respectively.It was possible to attribute structure 10e to the product obtained form 4e based on the observed relative change in chemical shifts of the aromatic hydrogen atoms upon acetylation of the free hydroxyl 18 .This chemical correlation is confirmed by previous results of Goodman and coworkers on the reaction of 4c,d with AlCl 3 6 .The regioselectivities observed in the methylenedioxyring opening step of the reaction of 4b-f with AlCl 3 allowed us to classify these substrates in three groups: (a) those which have only one strong electron-withdrawing substituent in the benzodioxole nucleus, i.e. 4b (CHO), 4d (CHO), and 4e (NO 2 ), and produce only one regioisomer of the chloro methyl ether intermediate (those with the chloromethoxy group located meta in relation to the electron-withdrawing substituent, 5b, 5d, and 6e, respectively); (b) those which contain no such groups, 4f and 4g, and produce an almost equimolecular amount of regioisomers 5f and 6f, and finally (c) 4c, that contains two such groups (CHO and NO 2 ) and, surprisingly 20 , produces only one isomer in which the chloromethoxy group is located meta in relation to the formyl substituent (5c).
Recent studies on the asymmetric opening reactions of chiral cyclic acetals have shown that site selective complexation by Lewis acid of one of the two oxygen atoms of these acetals plays a fundamental role in such reactions 21,22 .However, such a simple approach does not seem as reasonable for justifying the regioselectivities observed in the reactions of 1,3-benzodioxole nuclei as for those of chiral alicyclic acetals.In the case of compounds 4b-f other basic sites are available for complexation by AlCl 3 and our semiempirical (MNDO) calculations indicate that these coordinations would preferentially occur at the carbonyl groups 23 , which is in agreement with experimental Lewis basicity scales 24 .However, in none of the calculated carbonyl complexes (Figure 2) are the acetalic C-O bond lengths significantly altered with respect to the corresponding noncomplexed substrates 22 .So we propose that this kind of carbonyl complexation is not sufficient for activating the cleavage of the methylenedioxy ring.Two alternatives are possible: (a) a second complexation by AlCl 3 of one of the acetalic atoms would be necessary to promote the reaction, or (b) the methylenedioxy ring cleavage would occur via one of the less stable monocoordinated complexes.
The first of these assumptions is corroborated by experimental data: an excess of AlCl 3 is essential to convert 4b-g into the corresponding chloro methyl ether derivatives.The results of our calculations on the 1:2 complexes of 4b-d,f with AlCl 3 (Figure 3) indicate that the second complexation occurring at the oxygen atoms of the methylenedioxy ring activates the substrates for the cleaving process (significant changes of the acetalic C-O  bond lengths of 18,19, 20, and 21; Figure 3) 25 .Our interpretation is that the product forming selectivity should be controlled in the cleaving step itself of the activated dicoordinated (18/19 and 20/21, Figure 3) complexes 26 .When the possible transition structures for the reaction of the dicoordinated complexes of piperonal with AlCl 3 were calculated (Scheme 3), the barrier of activation for the reaction via the most stable complex (18b via 27) is higher (3.8 kcal mol -1 [0.91 kJ mol -1 ]) than that via the less stable one ( 19b via 28 ), showing a clear parallelism with the thermodynamic stabilities of the corresponding aluminum phenoxides (31 and 32, 4.5 kcal mol -1 [1.1 kJ mol -1 ]).Hence this assumed parallelism between barriers of activation and product thermodynamic stabilities was used to predict the reactivity of dicoordinated complexes.The relative energies of isomeric aluminum phenoxides 23-26 (Figure 4  From these results, we can conclude that a mechanism where the product forming selectivities are determined by the reaction rates of the dicoordinated complexes (18b-d vs 19bd) and not by their thermodynamic stabilities is operating in the methylenedioxy ring cleavage of 4b-d.In the case of 4f, we consider that the stability difference between 25 and 26 (Figure 4) is either overestimated at the semi-empirical theory level of our calculations or do not accurately reflects the difference in activation barriers for their reactions.However, considering that an excess of AlCl 3 is necessary for the reaction of 4f and that the difference in stability between 25 and 26 is the lowest so far calculated for all the dicoordinated complex products, we can also propose an analogous mechanism for the non-regioselective reaction of 4f, as well as for the regioselective reaction of 4e.

General
Melting points were measured with a Fisher-Johns (Fisher Scientific Co ) apparatus.Flash chromatography was performed using Merck silica gel 60, 230-400 mesh and tlc using Merck silica 60F 254 sheets . 1 H NMR and 13 C NMR were recorded on a Varian Gemini-200 instrument.Mass spectra were measured with a VG micromass spectrometer in the EI mode at 70 eV.The aluminum trichroride was commercially available and used whitout further purification.

General procedure for the preparation of 7b-d and 8e
To a solution of CH 2 N 2 (5.0 mmol) in Et 2 O (8.0 mL) at 0° C was added a catalytic amount of neutral Al 2 O 3 (0.01 mmol).After 5 min at 0° C a solution of 5 (1.0 mmol) in Et 2 O (2.0 mL) was added dropwise.The cold bath was removed and the resulting solution was allowed to gradually warm to room temperature and stirred for 4 h.The reaction mixture was concentrated in vacuum.Chromatography of the product on silica gel, using 4:1 hexane/EtOAc gave the procuct. 7b

Preparation of o-methoxyphenols 9b-d and 10e
To a solution of the 7 (1 mmol) in THF (10.0 mL) at room temperature, was added a 0.1 mol L -1 solution of AgNO 3 (5.0mL).The resulting mixture was stirred for 12 h.EtOAc (50 mL) was added and mixture was washed with brine, dried (Na 2 SO 4 ) and evaporated to dryness.Crude product was purified by flash chromatography to give 9b (70%) mp 112-114° C (lit. 10

Preparation of catechol 11b
To a suspension of AlCl 3 (5.0mmol) in CH 2 Cl 2 (4.0 mL) at room temperature, under N 2 , was added dropwise a solution of 4b (1 mmol) in CH 2 Cl 2 (5.0 mL).The resulting mixture was stirred for 3 h at room temperature.The mixture was cooled to 0ºC and cold water (0.04 mL) was added.The resulting mixture was stirred for 12 h at room temperature, under nitrogen.The reaction mixture was poured into brine solution and extracted with EtOAc (2 x 100 mL).The organic layer was washed with brine then dried (Na 2 SO 4 ) and concentrated under reduced pressure to give 11b in 81% yield; 1

Computational methods
In this study we have used the MNDO 7 Hamiltonian.MNDO, in spite of its well-known shortcomings 27 , has proven to be more reliable for studying aluminum-containing compounds than the more recent semi-empirical methods, AM1 and PM3 28 .
The calculations were performed in a IBM RS/6000, 25T workstation using MOPAC 6.0 29 program, and involved three steps.Firstly, the generation of relaxed potential energy maps for reactants (4b-d,f), their coordination complexes with AlCl 3 (4b-d,f.nAlCl3 ; n=1 to 3) and aluminum-phenolate products (29 to 32).These potential energy surfaces were determined by stepwise (10 to 30 o ) variation (reaction coordinate or grid index option) of all available acyclic torsion angles.During these grid searches all the geometrical parameters, except for the fixed pair of torsion angles being searched, were optimized under PRECISE and GNORM=0.5 key-words.Secondly, the local minima of these potential energy maps were determined under optimization of all variables with PRECISE and GNORM=0.01 key-words.When the default gradient norm minimization routine (BFGS) 30 of MOPAC was unable to attain a value less than 0.01 kcal -1 deg -1 , this could be easily obtained with the eigenvector following (EF) 31 routine.Finally, the transition structures (TS) for conversion of AlCl3-dicoordinated complexes of piperonal (18b and 19b) into the corresponding phenolate products (27 and 28) were located and refined.This was performed through the reaction coordinate method 32 , which involved stepwise variations of the C=C-C(H)=O torsion angles, and acetalic C-O or (Al)Cl-C(acetalic) distances.These initial guesses were then refined by a combination of NLLSQ 33 and TS 34 optimizers.
All the stationary points in these potential energy surfaces were characterized as minima or saddle points by vibrational frequency analysis.

Conclusions
A one pot procedure for the transformation of 4b,e into catechol derivatives 11b,e was developed.The mild conditions used in the hydrolyses of the chloro methyl ether intermediates and the isolation of pure catechols just by extraction of the aqueous layer represent an improvement of the reaction of AlCl 3 with safrole derivatives.The regioselective preparation of o-methoxyphenols 9b-d and 10e from the readily isolated chloromethylene derivatives 7b-d and 8e was described for the first time and increases the scope of the use of safrole derivatives in organic synthesis.Furthermore, these regioselective transformations have allowed us to unambiguously assign the structures of the chloro methyl ether intermediates.Based on these new experimental data and on semi-empirical (MNDO) molecular orbital calculations, we were able to propose a mechanistic rationale for the observed selectivities.