HETEROCYCLES

Paper | Special issue | Vol. 77, No. 1, 2009, pp. 461-469
Received, 10th July, 2008, Accepted, 28th August, 2008, Published online, 1st September, 2008.
DOI: 10.3987/COM-08-S(F)43

■ Reaction of β,β-Bis(trifluoroacetyl)vinyl Esters and β-Trifluoroacetylvinyl Esters with 1,2-Phenylenediamines Accessing Fluorine-containing Benzo[b][1,4]diazepine Derivatives — A Study about the Reaction Based on Molecular Orbital Calculations

Norio Ota, Yasuhiro Kamitori,* Takehisa Tomoda, Naoya Terai, and Etsuji Okada*

Department of Chemical Science and Engineering, Faculty of Engineering, Kobe University, Rokkodai-cho, Nada-ku, Kobe 657-8501, Japan

Abstract

β,β-Bis(trifluoroacetyl)vinyl ether (1) reacted with 1,2-phenylenediamine to give dihydrobenzodiazepinol (4a) selectively, whereas β-trifluoroacetylvinyl ether (2) and β-trifluoroacetyl-α-phenylvinyl ether (3) gave the corresponding O-N exchange products (6b, c) when reacted with 1,2-phenylenediamine. The factors determining the reaction products of the reaction of three substrates 1-3 having similar structures with 1,2-phenylenediamine were elucidated on the basis of molecular orbital calculations. The dehydration processes from dihydrobenzodiazepinols (4 and 8) to benzodiazepines (5 and 7) are also discussed.

INTRODUCTION
In recent years, much attention has been focused on the development of new methodologies for the syntheses of many kinds of fluorine-containing heterocycles, since these compounds are now widely recognized as important organic materials showing specific functions as well as interesting biological activities.1-4 In our previous paper,5 we reported an efficient and convenient synthetic method accessing fluorine-containing dihydrobenzo[b][1,4]diazepinols which have remarkable anti-tumor activities6 from β,β-bis(perfluoroalkanoyl)vinyl ethers. During the investigations, we found that the reaction of β,β- bis(trifluoroacetyl)vinyl iso-butyl ether (1) with 1,2-phenylenediamine gave 2,5-dihydro-3-
trifluoroacetyl-2-trifluoromethyl-1H-benzo[b][1,4]diazepin-2-ol (4a) selectively under very mild conditions without microwave irradiation. Our results showed clear contrast with Reddy’s reports of obtaining 1H-benzo[b][1,4]diazepine (5a) by the reaction of 1 with 1,2-phenylenediamine carried out under microwave irradiation.7,8 We also found that the reaction of β-trifluoroacetylvinyl iso-butyl ether (2) with 1,2-phenylenediamine produced only O-N exchange product (6b). A similar O-N exchange product was seen in Bonacorso’s work in which 6c was obtained as the sole product of the reaction of β-trifluoroacetyl-α-phenylvinyl methyl ether (3) with 1,2-phenylenediamine.9 Moreover, it has been reported that 6c was converted to the corresponding 3H-benzo[b][1,4]diazepine (7c) by heating 6c in the presence of acetic acid.9 In contrast, 6b did not give any benzodiazepines or dihydrobenzodiazepinols, even in the presence of acid catalyst.

The finding that the conversion of O-N exchange product 6c to benzodiazepine 7c occurred in the presence of acid catalyst suggests that dehydration on dihydrobenzodiazepinol 4c or 8c, which is thought to be the precursor of 7c, proceeded smoothly by acid catalysis. However, acid-catalyzed dehydration of dihydrobenzodiazepionol 4a was not successful and the corresponding benzodiazepine 5a was not obtained at all.5
We here present the most reasonable interpretation on the basis of molecular orbital calculations for these interesting differences in reactivity among three substrates 1-3 in the reaction with 1,2-phenylendiamine. Moreover, dehydration processes of dihydrobenzodiazepinols 4 and 8 to benzodiazepines 5 and 7 are also discussed.

RESULTS AND DISCUSSION
Derivatives of β-trifluoroacetylated vinyl ethers such as β-trifluoroacetylketene acetals,10 β,β-bis(trifluoroacetyl)vinyl ethers,11 β-trifluoroacetyl-α-phenylvinyl ethers,12 and β-trifluoroacetylvinyl ethers13 readily undergo nucleophilic O-N exchange reactions at olefinic carbons with various aliphatic and aromatic amines to give the corresponding β-trifluoroacetylated ketene O,N-acetals, β,β-bis(trifluoroacetyl)enamines, β-trifluoroacetyl-α-phenylenamines, and β-trifluoroacetylenamines. Consequently, O-N exchange products 6a-c were supposed to be the initial intermediates in the reaction of three substrates 1-3 with 1,2-phenylenediamine (Scheme 1). In the cases of β-trifluoroacetylvinyl iso-butyl ether (2) and β-trifluoroacetyl-α-phenylvinyl methyl ether (3), the reaction stops at this stage. In contrast, the subsequent intramolecular nucleophilic addition of the remaining aromatic NH2 group to trifluoroacetyl carbonyl group in 6 (6a) proceeds to give dihydrobenzodiazepinol 4 (4a) in the case of bis(trifluoroacetyl)vinyl iso-butyl ether (1) as illustrated in Scheme 2.

If dehydration of dihydrobenzodiazepinols 4 is possible, 1H-benzo[b][1,4]diazepines (5) is obtained. If isomeric 2,3-dihydro-1H-benzo[b][1,4]diazepin-2-ols (8) are thermodynamically more stable than 2,5-dihydro-1H-benzo[b][1,4]diazepin-2-ols (4), 3H-benzo[b][1,4]diazepines (7) are produced from 4 via 8.
As an initial step to clarify the reason why cyclization of O-N exchange products 6 to dihydrobenzodiazepinols 4 did not occur in the cases of 6b and 6c whereas it proceeded smoothly in the case of 6a, we computed the most stable structures of 6a-c and their energies (E€€6) using RB3LYP/6-31G*//RB3LYP/6-31G*. As depicted in Scheme 3, the transformation from the most stable conformers 6 to 9 suitable for subsequent cyclization would be necessary to convert 6 to dihydrobenzodiazepinols (4). We presumed that the facility of the transformation from 6 to 9 would be correlated with multiple bonding characters on C(1)-N(2) bond of 6 owing to the push-pull type canonical contribution of 6’. Thus, we calculated Mulliken bond orders14 on C(1)-N(2) bond of 6a-c and performed structural optimization for conformers (9a-c). In Table 1, the values of bond order PCN for 6a-c are listed together with the energies of 6a-c (E6) and 9a-c (E9).

PCN of 6a is apparently larger than those of 6b and 6c indicating enhanced multiple bonding character on C(1)-N(2) bond of 6a compared to those of 6b and 6c. Enhanced multiple bonding character would prevent the rotation around C(1)-N(2) bond on conformer (6a) and, consequently, the transformation from 6a to conformer (9a). Therefore, the above results are incompatible with the experimental results in which cyclization of intermediate (6a) occurred easily to give dihydrobenzodiazepinol (4a),5 while 6b and 6c did not cyclize to 4b and 4c, respectively. Therefore, the conformation change process from 6 to 9 is not important for the overall cyclization process from 6 to 4, showing that the cyclization process from conformers (9) to dihydrobenzodiazepinols (4) is a key step in determining whether O-N exchange products (6) are converted to 4.
We focused on intramolecular frontier orbital interactions, i.e. the interactions between nitrogen in aromatic NH2 group (HOMO) and carbonyl carbon in COCF3 group (LUMO) on conformers (9a-c). Thus, frontier electron densities, frHOMO at NH2 and frLUMO at COCF3 on 9a-c were calculated and the results are shown together with C-N distances between NH2and COCF3in Table 2. In the case of 9a bearing two trifluoroacetyl groups, frontier electron density on LUMO is concentrated at carbonyl carbon of the other trifluoroacetyl group, which is not a reaction center of the present cyclization reaction. Therefore, we used the electron density on the 2nd LUMO, which has a slightly higher energy level (ca. 0.65 eV) than LUMO on 9a.

Both values of frHOMO and frLUMO on 9a are apparently larger than those on 9b and 9c. These values of frontier electron density indicate that the intramolecular frontier orbital interaction on 9a would be considerably greater than those on 9b and 9c. In addition, the C-N distance on 9a being ca. 0.3 Å shorter than those on 9b and 9c would also assist the intramolecular frontier orbital interaction on 9a to a greater extent. The strong intramolecular frontier orbital interaction would promote the cyclization of 9a to dihydrobenzodiazepinol (4a) under very mild reaction conditions. In contrast, the intramolecular HOMO-LUMO interaction on 9b and 9c would not be strong enough to mediate cyclization of 9b and 9c to the corresponding dihydrobenzodiazepinols (4b and 4c), respectively, under similar reaction conditions.
To clarify the relative stability of dihydrobenzodiazepinols (4 and 8) depicted in Scheme 2, we computed the optimized structures of 4a,c and 8a,c together with their energies. Our results indicate that 8a is ca. 4.5 kcal/mol less stable than 4a, whereas 8c is ca. 10 kcal/mol more stable than 4c. Therefore, the isomerization process from 4a to 8a is thought to be inhibited. This would be a reason why the reaction of β-bis(trifluoroacetyl)vinyl iso-butyl ether (1) with 1,2-phenylendiamine produced dihydrobenzodiazepinol (4a) (not 8a) as a sole product. On the other hand, 4c is thought to isomerize immediately to more stable 8c when 4c is able to form by cyclization of 6c. However, it is necessary to take in account that acid catalyst was necessary for cyclization of 6c.9 As illustrated in Scheme 4, cyclization of 6c would proceed via cation (10) produced by protonation on carbonyl oxygen of conformer (9c) in the presence of acid catalyst to produce 2-hydroxy-2,5-dihydro-1H-benzo[b][1,4]diazepin-1-ium cation (11). Cation (11) was estimated to be 0.52 kcal/mol more stable than isomeric 2-hydroxy-2,3-dihydro-1H-benzo[b][1,4]diazepin-1-ium cation (12). Thus, isomerization from 11 to 12 would not be an energetically favorable process and, therefore, subsequent dehydration is thought to proceed predominantly from 11.

As for the dehydration processes from dihydrobenzodiazepinol (4a) to benzodiazepine (5a) in the presence of acid catalyst, we can postulate the most reasonable model reactions, as shown in Scheme 5. Cation (14) was employed as the most suitable precursor for acid-catalyzed dehydration of 4a. Dehydration of 14 gives 5H-benzo[b][1,4]diazepin-1-ium cation (15). On the other hand, cation (13) would be formed by dehydration of cation (11). Deprotonation of 13 gives 1H-benzo[b][1,4]diazepine (5c) which would isomerize to 3H-benzo[b][1,4]diazepine (7c) because 7c was estimated to be 6.1 kcal/mol more stable than 5c.
We compared the energies of cations (11 and 14) with the total energies of cations (13 and 15), respectively, and water. The process from 11 to 13 was predicted to be an exothermic reaction with a heat of reaction of -2.9 kcal/mol, while that from 14 to 15 was predicted to be endothermic with 10.4 kcal/mol. These results suggest that dehydration of cation (11) readily occurs to give 13, whereas that of cation (14) producing 15 in an energetically unfavorable process. Since cation (11) is produced by acid-catalyzed cyclization of 6c (Scheme 4) and deprotonation of cation (13), benzodiazepine (7c) is easily produced via 5c (Scheme 5), and the overall reaction from 6c to 7c is predicted to proceed smoothly by acid catalysis. However, acid-catalyzed dehydration of 4a to 5a via 14 is estimated to be difficult. These predictions are quite compatible with the experimental results5,9 where the O-N exchange product (6c) could be converted to the corresponding benzodiazepine (7c) by heating 6c in the presence of acetic acid, while acid-catalyzed dehydration of dihydrobenzodiazepionol (4a) to the corresponding benzodiazepine (5a) was unsuccessful.

We also carried out calculations about the dehydration process from dihydrobenzodiazepinol (4a) to benzodiazepine (5a) in the absence of an acid catalyst. The process from 4a to 5a and water was estimated to be an endothermic reaction with 17.8 kcal/mol (Scheme 5), suggesting that dehydration of 4a to 5a requires high external energy. It is likely that microwaves merely assisted the endothermic dehydration of dihydrobenzodiazepinols (4) to benzodiazepines (5) as an effective external energy in Reddy’s work7,8 because we found 4 could be readily obtained by the reaction of β,β-bis(trifluoroacetyl)vinyl iso-butyl ether (1) with 1,2-phenylenediamines without microwave irradiation under very mild conditions.5 In contrast with the case of 5c, 5a was estimated to be 3.6 kcal/mol more stable than 7a, which explains why the isomerization from 5a to 7a was not observed.7,8
Finally, we made calculations about the reaction of β,β-bis(pentafluoropropionyl)vinyl iso-butyl ether (16) with 1,2-phenylenediamine. This reaction was found to give a mixture of O-N exchange product (17) and dihydrobenzodiazepinol (19), and the complete conversion from 17 to 19 could not be achieved by prolonging reaction time (in Scheme 6).5

The cyclization process from conformer (18) to dihydrobenzodiazepinol (19) was estimated to be an endothermic reaction with a heat of reaction of 1.7 kcal/mol whereas that from conformer (9a) to dihydrobenzodiazepinol (4a) was predicted to be exothermic with -1.5 kcal/mol. In the case of the reaction of 16 with 1,2-phenylenediamine, the relative instability of 19 compared to 18 and possible equilibrium between 18 and 19 are thought to prevent the complete conversion from O-N exchange product (17) to dihydrobenzodiazepinol (19).

CONCLUSION
Based on molecular orbital calculations, we can rationalize the difference of products resulting from the reactions of β,β-bis(trifluoroacetyl)vinyl iso-butyl ether (1), β-trifluoroacetylvinyl iso-butyl ether (2), and β-trifluoroacetyl-α-phenylvinyl methyl ether (3) with 1,2-phenylenediamine. Intramolecular frontier orbital interaction on O-N exchange products (6) (conformers 9) as intermediates of the above reactions would be a key factor in determining whether the subsequent cyclization reactions yielding dihydrobenzodiazepinols (4) take place. Unsuccessful dehydration of dihydrobenzodiazepinol (4a) to benzodiazepine (5a) in the presence of acid catalyst is attributed to the endothermic dehydration process from protonated dihydrobenzodiazepinol (14) to protonated benzodiazepinol (15) requiring high energy.

COMPUTATIONAL METHODS
All calculations employed in this paper were accomplished using the computer programs packages SPARTAN and PC SPARTAN 04.15 All calculations for geometrical optimizations were performed with the 6-31G* basis set at B3LYP 16 level. The starting geometries employed for all optimizations were resulted from molecular mechanics using SYBYL17 force field and subsequent semi-empirical PM318 optimizations. The calculations for energy of intermediates were also taken with the 6-31G* basis set at B3LYP level.

This paper is dedicated to Professor Emeritus Keiichiro Fukumoto on the occasion of his 75th birthday.

References

1. R. Filler and Y. Kobayashi, ‘Biomedicinal Aspects of Fluorine Chemistry,’ Kodansha & Elsevier Biomedical, Tokyo, 1982.
2. R. Filler, ‘Organofluorine Chemicals and Their Industrial Applications,’ Ellis Horwood, London, 1979.
3. J. T. Welch, Tetrahedron, 1987, 43, 3123. CrossRef
4. R. Filler, Y. Kobayashi, and L. M. Yagupolskii, ‘Organofluorine Compounds in Medicinal Chemistry and Biomedical Applications,’ Elsevier, Amsterdam, 1993.
5. N Ota, T. Tomoda, N. Terai, Y. Kamitori, D. Shibata, M. Médebielle, and E. Okada, Heterocycles, 2008, 76, 1205. CrossRef
6. E. Okada, N. Ota, T. Tomoda, M. Fujimoto, and H. Takenaka, Jpn. Kokai Tokkyo Koho 2006-273844, 2006.
7. A. C. S. Reddy, P. S. Rao, and R. V. Venkataratnam, Tetrahedron Lett., 1996, 37, 2845. CrossRef
8. A. C. S. Reddy, P. S. Rao, and R. V. Venkataratnam, Tetrahedron, 1997, 53, 5847. CrossRef
9. H. B. Bonacorso, L. M. L. Marques, N. Zanatta, and M. A. P. Martins, Synth. Commun., 2002, 32, 3225. CrossRef
10. M. Hojo, R. Masuda, E. Okada, H. Yamamoto, K. Morimoto, and K. Okada, Synthesis, 1990, 195. CrossRef
11. M. Hojo, R. Masuda, E. Okada, and Y. Mochizuki, Synthesis, 1992, 455. CrossRef
12. M. Hojo, R. Masuda, and E. Okada, Synthesis, 1986, 1013. CrossRef
13. M. Hojo, R. Masuda, E. Okada, S. Sakaguchi, H. Narumiya, and K. Morimoto, Tetrahedron Lett., 1989, 30, 6173. CrossRef
14. R. S. Mulliken, J. Chem. Phys., 1955, 23, 1833; CrossRef R. S. Mulliken, J. Chem. Phys., 1955, 23, 1841; CrossRef R. S. Mulliken, J. Chem. Phys., 1955, 23, 2338; CrossRef R. S. Mulliken, J. Chem. Phys., 1955, 23, 2343. CrossRef
15. Wavefunction, Inc.
16. A. D. Becke, J. Chem. Phys., 1993, 98, 5648. CrossRef
17. M. Clark, R. D. Cramer III, and N. van Opdensch, J. Computational Chem., 1989, 10, 982. CrossRef
18. J. J. P. Stewart, J. Computer Aided Molecular Design, 1992, 6, 69