Novel dicyanomethylene compounds from the reaction of 2-aryl-2,3-dihydro-1 H -pyrrolo[3,4-b ]quinoxalines with ethenetetracarbonitrile

2-Aryl-2,3-dihydro-1 H -pyrrolo[3,4-b ]quinoxalines


Results and Discussion
Starting materials 8a-d were prepared from 2,3-bis(bromomethyl)quinoxaline and anilines.Upon addition of doubled molar amounts of TCNE to solutions of 8a-d in benzene with admission of air, a green colouration is observed which gradually gives way to a brown and finally to a characteristic blue colour.After concentration, the residue was subjected to vacuum sublimation to remove any unreacted TCNE.Chromatographic separation of the residue gave some tarry material and numerous coloured zones.From the three most intense zones products 9-11 could be isolated (Scheme 2).
Solutions of compounds 9a-d show a characteristic deep blue colour, attributable to the local push-pull systems of conjugated double bonds and lone pairs.The λ max values of 9a-d in the visible spectrum (586-602 nm in acetonitrile solution, log ε 4.12-4.45)are very close indicating the same molecular gross structure and configuration.They are also in good agreement with the visible spectra of nine compounds of type 2 (λ max 570-584 nm, log ε 4.33-4.84 in acetonitrile) reported earlier. 10he structures of 9a-d are further supported by characteristic IR, NMR and mass spectral data.For example, the 1 H-NMR spectrum of 9b in CDCl 3 clearly shows the absence of methylene protons but the presence of two isochronous methyl groups located on the two phenyl rings (one singlet only at δ 2.30 ppm for six protons), and phenyl protons.The gross formula C 40 H 22 N 10 for 9b was confirmed by the mass spectrum which exhibited the molecular ion at m/z 642 (31%).It should also be noted that the mass spectra of compounds 9a-d show the loss of C(CN) 2 from the molecular ions but the correspondent fragment peaks are of relatively low intensity.The IR spectrum of 9b shows a sharp absorption at 2215 cm -1 characteristic for conjugated C≡N groups.
The configurational assignment of compounds 9a-d needs to be adressed.In neither the Znor the E-isomers can the N-aryl group be coplanar with the adhering pyrroloquinoxaline unit.Therefore, in the E-form the "inner" quinoxaline N-atoms would show an infavourable interaction with the N-aryl π-systems, and the "inner" peri-protons would experience the shielding influence of the N-aryl ring.No significantly upfield shifted aryl-H signals are detectable, however.Therefore, the Z geometry is proposed as most likely for compounds 9a-d in accordance with their close structural relationship to the clearly cisoid {3- , the structure of which had been confirmed unambiguously by a single crystal X-ray structure analysis.

Scheme 2
Similarly, the structures of products 10a-d are supported by their analytical and spectral data.For example, the elemental analysis of 10b suggested a gross formula C 27 H 18 N 6 .This was confirmed by the mass spectrum which exhibited a molecular ion at m/z 426 (100%).The IR spectrum showed a strong band at 2214 cm -1 due to conjugated cyano groups.The 300 MHz 1 H-NMR spectrum also clearly indicates the absence of any signals at δ 4.83 ppm expected for CH 2 groups of the 1,3-dihydropyrrole ring, but shows two singlets at δ 2.30 and 2.40 ppm with a 1:1 ratio assignable to the two different methyl groups, in addition to the aromatic proton signals.
The structure of quinoxaline-2,3-diol (11) was assigned on the basis of its spectroscopic properties (see experimental part) and by comparison of its melting point with the literature value. 14he reaction conditions provide an overall dehydrogenating and oxygenating environ-ment.For the formation of the main products 9a-d, an initial didehydrogenation of the starting materials 8a-d to the fully unsaturated 2H-pyrrolo [3,4-b]quinoxalines 12 seems to be most logical.This may, by analogy to the reactions of TCNE with isoindolines 1 as reported earlier, 10 be effected either by a concerted hydride transfer 15 from C1 of 8 to TCNE followed by proton transfer to generate 12 together with 1,1,2,2-tetracyanoethane (TCNE-H 2 ), or by a consecutive (non-concerted) sequence of electron and proton transfers leading to the same products.While the colouration effects observed immediately after combination of the reactant solutions do point to the formation of charge-transfer complexes, they do not allow a clear decision in favour of the non-concerted pathway for didehydrogenation of 8 to 12.For this reaction to occur via radical ion pairs an additional electronic excitation may be required 16 , which is not available under the conditions used in this study.Intermediates 12 are not isolated and very likely in turn react with TCNE to form tetracyanoethylated products 13 (Scheme 3), while there is so far no indication of formation of Diels-Alder type cycloadducts.The conversion 12 → 13 is in fact the first step of a so-called tricyanovinylation. 17Instead of expelling one molecule of hydrogen cyanide, however, two molecules of 13 react (in analogy to the case reported before 10 ) to form Z-9 via 14 with successive release of two molecules of malononitrile (Scheme 3).This sequence is a dehydrogenative symmetric coupling and at the same time an α-dicyanomethylenation.A closely analogous introduction of a dicyanomethylene group has been observed in the reaction of 1hydroxy-2-phenylindole with TCNE yielding (1-oxido-2-phenyl-3H-indol-3ylidene)propanedinitrile. 18In contrast to the case reported here, 2-unsubstituted isoindoles (as 1phenylisoindole) seem to undergo oxidative bimolecular couplings via their radical cations. 19,20he formation of products 10a-d also clearly involves the conversion of two molecules of starting material 8a-d, one being dehydrogenated to 12 and further converted into 13, the other being degraded to provide a source of the (N-aryl)imino function.An electrophilic aryl-nitogen species attacking C3 of 13 would most logically serve this purpose.The nature of this species, however, has not been unraveled yet.An arylnitrene 18, as proposed in Scheme 4, is just one possibility.
Several reports of either cheletropic or stepwise release of nitrogen species from various types of 1,4-dihydro-1,4-iminonaphthalenes or higher anellated anlalogues thereof are found in the literature, the prime motive of which is removal of the imino bridge to generate anellated arenes.In all cases, prior to extrusion, the imino nitrogen atom is attached to an electrophilic atom or group, such as oxygen from a peracid, 21,22 nitrosyl, 23 dichlorocarbene, 24 amino (followed by dehydrogenation), 25,26 but also dimethyl ethyne dicarboxylate 27 and benzyne 27 .It had been suggested 27,28 that the latter two form zwitterionic adducts from 1,4-iminonaphthalenes by C-N bond heterolysis, and that these zwitterions either release the naphthalene or undergo cyclization to 1,2-heteroanellated naphthalenes.
Thus, TCNE could, by analogy to the abovementioned behaviour of dimethyl ethynedicarboxylate, 27 add to N2 of 8, and the resulting zwitterion 15 could release 16 and the betaine 17 (Scheme 4).Either 17 or the nitrene 18 released therefrom could serve as electrophilic agent on 13.With 18 (in a prototypical sense) intermediate 19 would be formed, which in two steps would afford 10a-d with release of malononitrile.The betaine 17 could act in the same sense.Also, the radical cation formed from 8 and TCNE could, in principle, release the nitrene 18, and in all cases products 10a-d would finally be reached.A similar pathway had been proposed earlier for the formation of compounds 3 from isoindolines 1. 13 However, no products derived in any way from 16 have been identified so far.

Scheme 4
Last, byproduct 11 is probably formed by a Diels-Alder type addition of oxygen to 12 followed by rearrangement of the primary adduct via 20 to 21 (not isolated).The latter compound, being sensitive to water, can generate 11 with the likely release of glyoxal mono-(Naryl)imine (Scheme 5).Isoindoles are sensitive to oxygen, 19 and formation of endo-peroxides which may undergo homolysis to bis-alkoxyl biradicals and further reactions thereof has been reported for a series of N-substituted 1,3,4,7-tetramethylisoindoles. 29hen evaluating the yields it needs to be considered that the formation of both products 9 and 10 each requires the transformation of two molecules of 8. Since the reactions occurring involve numerous steps and unidentifiable tarry products are also formed, moderate yields have to be accepted.

Conclusions
As previously reported for the transformations of isoindolines 1 to products 2 and 3, 10,13 dicyanomethylenation and overall dehydrogenative symmetric couplings of the title compounds 8a-d under the action of ethenetetracarbonitrile (TCNE) have been observed and plausibly rationalized on the basis of relevant findings taken from the literature.The results reported here supplement the rich chemisty of TCNE. 17Aza-substitution in the starting materials 8a-d did not lead to decreased reactivity towards TCNE but rather activated their methylene hydrogen atoms.

Experimental Section
General Procedures.The uncorrected melting points were determined with a Reichert Thermovar hot stage microscope.The IR spectra were obtained from KBr disks on either a Shimadzu 408 or a Bruker Vector 22 FT-IR instrument.The 300 MHz 1 H-NMR spectra were recorded on a Bruker WM 300 instrument with tetramethylsilane as internal reference, s = singlet, d = doublet, m = multiplet.The mass spectra (70 eV, electron impact mode) were obtained on an AMD 604 instrument.The visible spectra were recorded on a Perkin-Elmer Lambda 2 spectrophotometer.Combustion analyses were run on a Carlo Erba Model 1106 CHN analyser, the results were found to be in good agreement (±0.2%) with the calculated values.Preparative layer chromatography was carried out using air dried 1.0 mm thick layers of slurry applied silica gel Merck PF254 on 48 cm wide and 20 cm high glass plates.Zones were detected by their colour or by quenching of indicator fluorescence upon exposure to 254 nm light.

Preparation of 2-aryl-2,3-dihydro-1H-pyrrolo[3,4-b]quinoxalines (8a-d).
To a solution of 3.16 g (10 mmol) of 2,3-bis(bromomethyl)quinoxaline in 50 ml of dry toluene, 10 mmol of arylamine and 2.02 g (20 mmol) of triethylamine were added.The mixture was kept under reflux for 3h.After cooling to room temperature, the reaction mixture was filtered and the precipitate was washed thoroughly with chloroform.The washings were shaken with water, dried over anhydrous calcium chloride, and concentrated to give a yellow precipitate, which was crystallized from toluene.

Reaction of 2-Aryl-2,3-dihydro-1H-pyrrolo[3,4-b]quinoxalines 8a-d with TCNE.
A solu-tion of 2.0 mmol of 8a-d in 30 ml of dry benzene was added dropwise with stirring to 512 mg (4.0 mmol) of TCNE in 40 ml of benzene at room temperature.Stirring was continued for 7 days with admission of air.The colour of the mixture gradually changed from green to brown and finally gave way to a blue colour.The reaction mixture was concentrated to dryness and the residue subjected to sublimation at 80 o C under vacuum to remove all unreacted TCNE.The residue was then separated by preparative layer chromatography using cyclohexane/ethyl acetate 3:1 as eluent to give numerous coloured zones, the three most intense of which were removed and extracted with acetone.The concentration residue was rechromatographed with the same eluent to enhance separation.The fastest migrating zone, which quenched all indi-cator fluorescence upon exposure to 254 nm UV-light, contained quinoxaline-2,3-diol (11), the second zone (orange colour) contained compounds 10a-d, whereas the most slowly migrating zone (which is always characterized by a deep blue colour) contained compounds 9a-d.The dried extracts were crystallized from suitable solvents.