Synthesis of quinoxalines by cyclization of α-arylimino oximes of α-dicarbonyl compounds

Article abstract of DOI:10.1016/S0040-4039(00)01847-5

Heating the title compounds 1 at reflux in acetic anhydride yields quinoxalines 3 and 4 via a presumed aryliminoiminyl radical 5, resulting from homolytic cleavage of the N-O bond in the intermediate ester 2. The observed regioselectivity of the reaction is also rationalized by implicating such a radical. (C) 2000 Elsevier Science Ltd.

Full text of DOI:10.1016/S0040-4039(00)01847-5

TETRAHEDRON
LETTERS
Pergamon
Tetrahedron Letters 41 (2000) 10299–10302
Synthesis of quinoxalines by cyclization of a-arylimino
oximes of a-dicarbonyl compounds
N. P. Xekoukoulotakis, C. P. Hadjiantoniou-Maroulis and A. J. Maroulis*
Department of Chemistry, Aristotle University of Thessaloniki, PO Box 103, GR-540 06 Thessaloniki, Greece
Received 30 June 2000; revised 5 October 2000; accepted 19 October 2000
Abstract
Heating the title compounds 1 at reflux in acetic anhydride yields quinoxalines 3 and 4 via a presumed
aryliminoiminyl radical 5, resulting from homolytic cleavage of the NꢀO bond in the intermediate ester 2.
The observed regioselectivity of the reaction is also rationalized by implicating such a radical. © 2000
Elsevier Science Ltd. All rights reserved.
Keywords: a-arylimino oximes; iminyl radical; quinoxalines; spirodienyl radical.
Quinoxaline derivatives are very useful compounds with well-known biological activity.1,2 Of
the routes available for the synthesis of quinoxalines, the cyclization of appropriately function-
alized 1-N-4-aryl-1,4-diaza-1,3-dienes is the most promising in terms of regiochemistry and
generality. However, the first of two early examples, which involved cyclization of a-arylimino
phenylhydrazones of a-dicarbonyl compounds, employed extremely vigorous conditions and was
plagued by low yields and other practical disadvantages.3 The second, which dealt with the
photocyclization of acetate and benzoate esters of the a-phenylimino oxime of benzil, was rather
limited in scope and also suffered from low yields.4 We recently reported an oxidative cyclization
of benzil a-arylimino oximes 1 (R1=R2=Ph) to 2,3-diphenylquinoxaline 1-oxides as another
representative example of this approach.5
As previously mentioned, what makes the cyclization of the diaza-dienes 1 to quinoxaline
derivatives so promising is the expected regioselectivity in those cases where nonsymmetrical
substrates are used. The lack of regioselectivity is the main drawback of the most widely used
method for quinoxaline synthesis, namely the condensation of o-phenylenediamines with
a-dicarbonyl compounds.1,2
Our investigations on the preparation of the title compounds initially involved the synthesis,
isolation and thermal conversion of the oxime acetates 2 to the quinoxalines 3 (Scheme 1 and
* Corresponding author. Fax: 03031-997679; e-mail: apm@chem.auth.gr
0040-4039/00/$ - see front matter © 2000 Elsevier Science Ltd. All rights reserved.
PII: S0040-4039(00)01847-5

10300
R3
R3
R4
R1
R4
R1
N OH
R2
N OAc
R2
Ac2O, reflux
N
N
1
2
R3
R3
R4
R1
R4
N
R1
R2
N
N
+
R2
N
3
4
Scheme 1.
Table 1, entries 2–4 and 8). However, we soon realized that the one pot procedure described
here, which employs the oximes 1 instead of their esters 2, provided comparable or better yields
of quinoxalines under especially mild conditions in only one step.6
When the starting a-arylimino oximes 1 originating from symmetrical diketones (R1=R2)5
and when the aryl group bore only a para substituent (R3=H), this substituent ended up
naturally in the 6-position of the quinoxaline system (i.e. 3=4 in this case, Table 1, entries 2–4).
Interesting site selection was observed, however, when the meta substituted derivatives 1g–i
and 2h were cyclized. Contrary to that normally expected on the basis of steric considerations,
the regioisomers 3g–j, resulting from attack at the position ortho to the R3 subsituent, prevailed
over regioisomers 4g–j originating from attack at the para position. It is clear from the results
shown in Table 1 (entries 7–9) that the ratio 3/4 is only slightly dependent on the substituent (in
all of these cases R1=R2).
Table 1
Isolated quinoxalines from a-arylimino oximes 1 and their corresponding esters 2
Entry Starting oxime (ester) 1 (2)
R1 R2 R3
R4
Quinoxaline yield (%)
3
4
Total 3+4
Ratio 3/4
1
2
3
4
5
a
b
c
d
e
f
Ph Ph
Ph Ph
Ph Ph
Ph Ph
Me Ph
Me Me
H
H
H
H
H
H
H
57
MeO 65 (64)a
Me
Cl
H
63 (62)a
53 (82)a
66
6
H
16
7
8
9
10
11
12
13
g
h
i
j
k
l
Ph Ph MeO
Ph Ph Me
Ph Ph Cl
Ph Ph Me
H
H
H
Me
47
32
79
1.5
51 (51)a 25 (22)a 76 (73)a
2.0 (2.3)a
48
47
26
29
74
76
1.8
1.6
Me Ph
Me Ph
Me Ph
H
H
H
MeO 50 [16]b 17 [6]b
67 [22]b
62
2.9 [2.7]b
Me
Cl
44
36
18
18
2.4
2
m
54
a Starting from the corresponding ester 2 after reflux in xylene.
b After reflux in Ni(dust)/Ac2O, AcOH.9

10301
The slight preference for ortho substitution and the small substituent effect on the yields
observed are in accord with homolytic aromatic substitution by iminyl radicals depicted in
Scheme 2 (vide infra). The ortho substituted product is often dominant in intermolecular
homolytic aromatic substitution reactions.7
R4
R4
N
N
Me
Ph
Me
Ph
N
N
1k-m
2k-m
5k-m
4k-m
R3
R1
R2
Me
Ph
N
N
R4
R4
N
N
OAc
6k-m
8
N
N
Me
Ph
N
Me
Ph
R4
R4
N
7k-m
Scheme 2.
5k-m
When the starting diketone was unsymmetrical (R1=Me, R2=Ph), the corresponding a-
arylimino oximes 1 (Table 1, entries 11–13, R3=H), although substituted exclusively in the para
position of the arylimino group, yielded both regioisomers 3k–m and 4k–m.
We believe that cyclization of the a-arylimino oximes 1 to the corresponding quinoxalines, in
acetic anhydride, proceeds, after an initial acetylation step, via the iminyl radical 5 and the
spirodienyl one 6. This is exemplified in Scheme 2, for R1=Me, R2=Ph and R3=H. The
degenerate rearrangement of radical 5 to the isomeric radical 7 via the spirodienyl radical 6 was
not unexpected since a similar rearrangement was reported by McNab two decades ago.8 This
rearrangement explains the formation of both regioisomers 3 and 4 (Table 1, entries 11–13) and
reflects the ability of radicals to attack ipso positions.7
In order to verify our proposition we performed the cyclization of 1k with nickel powder in
a mixture of acetic anhydride and acetic acid. These conditions were considered appropriate for
the generation of iminyl radicals from ketoximes.9 We obtained the same product distribution
(Table 1, entry 11) but in substantially lower overall yield. The above result provided further
evidence for the operation of the proposed mechanism and particularly for the intermediacy of
the iminyl radical 5.
Since the cyclization of a hetero-1,3,5-hexatriene to a five-membered ring is possible in
principle, one has also to consider pericyclic or intramolecular nucleophilic addition routes to
generate an intermediate, such as 8.10 Intermediate 8, after acetate equilibration and reopening,

10302
could lead to a regioisomer of the starting oxime ester 2 and ultimately to both regioisomers 3
and 4. However, such a situation was not encountered recently in the closely analogous case of
the electrocyclization of the carbonates of 1-hydroxy-1,4-diaza hexatrienes which resulted in
only one of the two possible regioisomeric pyrazine products in good yields.11
In spite of the above evidence it is still possible that a portion of the reaction follows a
non-radical route. The small difference observed in product distribution when the leaving group
was changed from acetyloxy (3h/4h=2.3) to t-butoyloxy (3h/4h=1.4) and our failure to detect
products resulting from the presumed decomposition of the acyloxy radicals might be suggestive
to this effect.
In conclusion, the results now reported represent a simple, mild and straightforward method
for the synthesis of quinoxaline derivatives via iminyl radicals by the use of easily available
precursors and cheap reagents. It suffices to say that the lack of mild processes for the
generation of iminyl radicals was considered the main reason for the scarce, until recently,
published work on iminyl radicals.12
References
1. Cheeseman, G. W. H.; Werstiuk, E. S. G. Adv. Heterocycl. Chem. 1978, 22, 367–431.
2. Sato, N. In Comprehensive Heterocyclic Chemistry II; Katritzky, A. R.; Rees, C. W.; Scriven, E. F. V., Eds.;
Pergamon: Oxford, 1996; Vol. 6, pp. 233–278.
3. McNab, H. J. Chem. Soc., Perkin Trans. 1 1982, 1941–1945.
4. Armesto, D.; Horspool, W. M.; Apoita, M.; Gallego, M. G.; Ramos, A. J. Chem. Soc., Perkin Trans. 1 1989,
2035–2038.
5. Maroulis, A. J.; Domzaridou, K. C.; Hadjiantoniou-Maroulis, C. P. Synthesis 1998, 1769–1772.
6. Typical experimental procedure: A 1 mmol solution of the a-arylimino oxime 1 in 5 mL of acetic anhydride was
heated at reflux. The progress of the reaction was monitored by TLC. At the early stages of the reaction the spot
corresponding to the oxime 1 disappeared and a new spot appeared corresponding to the oxime acetyl ester 2.
The reflux was continued until the oxime ester 2 was fully consumed (usually 8–24 h). The solvent was removed
in vacuo and the residue was column chromatographed (low pressure, silica gel, mixtures of petroleum ether/ethyl
acetate of increasing polarity). All compounds were identified by the usual spectroscopic techniques and gave
satisfactory elemental analyses. In particular, compounds 3k–m and 4k–m were identified by converting the
former product to the corresponding mono N-oxides.
7. Gandour, R. D. Tetrahedron 1980, 36, 1001–1009 and references cited therein.
8. McNab, H. J. Chem. Soc., Chem. Commun. 1980, 422–423.
9. Boivin, J.; Schiano, A.-M.; Zard, S. Z. Tetrahedron Lett. 1992, 33, 7849–7852.
10. George, M. V.; Mitra, A.; Sukumaran, K. B. Angew. Chem., Int. Ed. Engl. 1980, 19, 937–983.
11. Bu¨chi, G.; Galindo, J. J. Org. Chem. 1991, 56, 2605–2606.
12. Zard, S. Z. Synlett 1996, 1148–1154.
.