2. Results and Discussion

2.1 The Thiolactone Precursor 35

The generation of 8,8­dimethyl isobenzofulvene 10 in the presence of thiophosgene 34 afforded, in excellent yield, the adduct 35 whose structure was readily confirmed by mass spectrometry and p.m.r. The mass spectrum shows a molecular ion at m/e 270 with an isotope pattern characteristic of a dichlorinated compound [23]. The p.m.r. spectrum (Figure 1) reflects the different environments of the bridgehead protons, Ha and Hb, and the methyl groups. Assignment of the bridgehead resonances was made by comparison with the p.m.r. spectra of the other heterocyclic compounds 37, 39 and 25.

Figure 1

The adduct 35 is moisture sensitive and on exposure to the atmosphere rapidly rearranges to a bright yellow crystalline compound which was assigned the structure 42. The mass spectrum of this compound shows a molecular ion at m/e 234 with an isotope pattern characteristic of a monochlorinated compound [23]. The p.m.r. spectrum (Figure 2) shows only one high field signal, indicating the similar environment of both methyl groups. The loss of the methano bridge was indicated by the absence of resonances in the region d 3.0-5.0, and the appearance of the signal at d 6.25 suggested the presence of an olefinic bond. Attempts to purify the adduct 35 by chromatrography on silica gel or alumina yielded only the rearranged compound 42.

Figure 2

At this stage, two mechanisms for this rearrangement were proposed (Scheme 8).

Scheme 8

Path A involves the initial formation of a "tight" ion pair 40, subsequent bonding to the benzylic position (pseudo 1,3­migration) to form the dichloro compound 41 and elimination of the HC1 from this intermediate to yield the chloro compound 42. Path B involves a mechanism similar to that proposed for the ring-opening of the diazo compound 12 to form 13 to yield the indene intermediate 43. Subsequent acid-catalysed cyclization of this thiol again gives the intermediate dichloro compound 41 which loses HC1 to yield the observed product 42.

In an effort to determine whether the mechanism of this rearrangement followed Path A or Path B, the diphenyl substituted compound 46 was synthesized (Scheme 9). If the rearrangement followed Path B, then no rearrangement of the diphenyl adduct 46, should be observed. However, when set aside under atmospheric conditions, the diphenyl adduct 46 also rearranged to give the corresponding diphenyl compound 47 (Scheme 9). Compound 47 was identified by its mass spectrum, molecular ion at m/e 358 showing the isotope pattern of a monochloronated compound, and its p.m.r. spectrum (Figure 3), which shows a vinylic proton signal at d 6.44. Although it was possible that compounds 35 and 46 were undergoing rearrangement by different mechanisms, these observations suggest that both compounds rearrange via Path A (Scheme 8).

Figure 3

Scheme 9

The propensity towards rearrangement exhibited by the adduct 35 posed serious problems. Numerous attempts to effect the hydrolysis of 35 were made but only in aqueous ethanol were products other than 42 obtained. In this case, two additional compounds were obtained. These proved to be isomeric and their mass spectra exhibited a molecular ion m/e 216, which corresponds to a molecular formula C13H12SO. Both isomers crystallized readily from ethanol, isomer I as clear, bright yellow coloured crystals and isomer II as clear, pale brown coloured crystals, this colour difference being reflected in their uv/visible spectra. The ir spectrum of isomer II showed a strong band at 1695 cm-1 (C=O stretch) which was not apparent in the spectrum of isomer I. The p.m.r. spectra of the two isomers (Figure 4 and Figure 5) were very similar. On the basis of these data, structure 48 was assigned to isomer II and 49 to isomer I (Scheme 10).

Figure 4

Figure 5

Scheme 10

Confirmation of the assignments was obtained from their 13C nmr spectra and from the p.m.r. and uv spectra of their enolate anions. The proton decoupled 13C nmr spectra of the two isomers are illustrated in Figure 6 and Figure 7. The presence of the thione groups in isomer I and ketone in isomer II was confirmed by the presence of the low­field signals assigned to C3 in each spectrum [24]. Reference to the proton coupled spectra (not illustrated) allows the preliminary assignment of secondary, tertiary and quaternary resonances. By compiling the chemical shift correlation diagram illustrated in Figure 8, and by reference to previously reported 13C nmr studies [25] most of the signals in the spectra can be assigned.

Figure 6

Figure 7

Figure 8

The assignments of the methyl and methylene carbons, C9, C10 & C8, are readily made on reference to proton coupled spectra. C1, the only saturated quaternary carbon is also easily assigned to the high­field quaternary signal as illustrated. The remaining signals are assigned on the basis of their observed shift when the spectra of the two isomers are compared. The signal undergoing the greatest downfield shift is assigned to C3a, the carbon adjacent to C3. The greater g­effect of the thione function results in an upfield shift of the resonances of the two carbons g to C3. The two signals which show an upfield shift are thus assigned to C8a and C3b. The lower field signal is assigned to C8a on the basis of its shorter relaxation time (more intense signal) [26]. C7a is thus assigned to the remaining quaternary resonance. The remaining signal which undergoes a shift is assigned to C4. The other three aromatic signals cannot be assigned with certainty, but following the trend that the C8a and C7a signals appear downfield from the C3a and C3b signals respectively, C7, C6 and C5 are tentatively assigned as shown.

Both p.m.r. and uv spectral data indicate, that in basic media (-OH/acetone-d6 and -OH/95% ethanol respectively), the thione and ketone form the enolate anions illustrated in Figure 9. (Compare with Figure 2 & Figure 3). In basic ethanol, the uv of the two isomers show maxima at l 240, 270 & 380. (Compare with uv data for compounds 42 and 47).

Figure 9

A subsequent investigation into the origin of compounds 48 and 49 suggested that two modes of formation can be distinguished. In the first of these isobenzofulvene 10, reacts with thiophosgene in ethanol to yield ultimately, the enol ether 50. The details of the participation of the ethanol in this reaction are not clear, as thiophosgene reacts readily with ethanol and it is possible that the active dienophile may be ethyl chlorothioformate (to form 50a) (Scheme 11) as thioesters are recognised dienophiles [26]. The enol ether 50 will readily hydrolyse to the ketone 48, and has been observed to do so simply on exposure to moisture.

Scheme 11

Scheme 12

The second route leading to the isomers 48 and 49 has been shown to involve the thiophosgene adduct 35 (Scheme 12). In the presence of aqueous ethanol the adduct 35 rearranges to both 48 and 49, possible via a zwitterionic species such as 51. It is interesting to note that a variation of pH significantly effects the ratio of the isomers 48 and 49 formed. In some experiments, traces of the enol ether 50 was observed to be present with both 48 and 49. It appears possible then, that the proposed mechanisms may in fact compete with one another. If this is so, the variation of the product ratio 48:49 with pH may simply be due to one mechanism dominating the other.

2.2 The Lactone Precursor 37

When 8,8­dimethylisobenzofulvene 10 was generated in the presence of diethyl oxomalonate 36, a Diels­Alder addition resulted, to yield the adduct 37. This could not be crystallized from the crude mixture and, again, attempts to purify the crude product by chromatography on silica gel or alumina only resulted in its rearrangement. The p.m.r. spectrum of the crude product showed the presence of adduct 37: signals at d 4.74 & 5.76 for the bridgehead protons were definitive.

When chromatographed on silica gel, or simply when set aside, the crude diethyl oxomalonate adduct 37 rearranged to give two new products. Both structures showed the same molecular ion at m/e 330, and their mass spectra were almost identical. However, the structures of the two compounds were easily deduced as 53 and 54 (Scheme 13), from the p.m.r. spectra, Figure 10 and Figure 11, respectively.

Figure 10

Figure 11

Scheme 13

Deuterium exchange observed on addition of D2O to a solution of the major product 52 (Figure 12), revealed the presence of a hydroxyl function. The low-field resonance at d 6.84 (1H), by analogy with the sulphur compounds 42 and 47, suggested the presence of the indene nucleus. The presence of only one methyl proton signal (3H), other than the triplets of the ethoxy-group-methyl protons, and the presence of the resonance at d 5.08 (2H), confirmed the assignment of structure 54 to the major product. The resonance at d 4.76 was consistent with the presence of the methine proton Hb.

Figure 12

The p.m.r. spectrum of the minor product also shows a low­field resonance at d 6.58, again suggesting the presence of the indene nucleus. In this molecule however, this signal shows a coupling of 1.5 Hz to the resonance at d 5.00. By analogy with the alcohol 54, and the extent of this coupling, the resonance at d 5.00 is assigned to a benzylic methine proton. The fact that coupling exists, indicates the planar arrangement of the two benzylic protons suggesting that the indene 5­membered ring is puckered. Since the other signals in the spectrum are consistent with a non­planar structure, the structure 53 was assigned to the minor product. This structure was confirmed by the observation that in acidic, neutral or basic aqueous solution, the compound 54 was irreversibly converted to the alcohol 54.

These results suggested that the rearrangement of the adduct 37 proceeded via the mechanism outlined in Scheme 14 , or an acid-catalysed variant.

Scheme 14

When the crude diethyl oxomalonate adduct 37 was chromatographed on a column of deactivated basic-alumina, only one rearranged compound was eluted. The spectral data of this compound showed that it was not one of the previously observed compounds, 53 or 54. The mass spectrum showed a very weak molecular ion at m/e 348. The p.m.r. spectrum (Figure 12), when a solution of this compound was treated with D2O, indicated the presence of two hydroxyl functions. The coupling of 10 Hz between the higher field hydroxyl signal and the resonance at d 5.38 (1H) could only occur if the two functions were bound to the same carbon atom. The appearance of the only other single proton resonance at d 4.9 gave reasonable confidence in the assignment of structure 52 to this product. The syn configuration is tentatively assigned to this structure on the basis of a "W"-coupling of 1 Hz observed between Ha and Hd, which would not be observed if the diol 52 had the anti configuration.

Figure 13

The diol 52 is considered to be derived from the addition of water to a species such as 55. It is significant to note that in the current study, the diol 52 is the only compound isolated, to date, in which an external nucleophile has attacked a benzylic carbon atom. This must result from a heterogeneous reaction occurring on the alumina surface. However, the mechanism of this hydration is obscure and has not been pursued further.

The reaction of 7-diphenylmethylidene benzonorbornadiene 44 with s-tetrazine [27] in the presence of diethyl oxomalonate 36 gave no recognizable products.

2.3 The Lactam Precursor 25

Hydrolysis of the tosylate 39, synthesized by the Diels-Alder addition of tosyl cyanide 38 [3] to 8,8­dimethyl isobenzofulvene 10, in acetic acid, afforded the b­lactam 25 (Scheme 16). The structure of the lactam was readily assigned on the basis of its pmr spectrum (Figure 13) and its mass spectrum (Figure 14). In contrast to the lactone and thiolactone adducts 35 and 37 the lactam 25 did not rearrange on silica gel or alumina, and readily dissolved in glacial acetic acid without rearrangement.

Scheme 15

Photolysis of the Lactam 25

Preliminary photolyses of the b­lactam have been carried out. Purification of the crude photolysis mixture by chromatography on silica gel, however, yielded no hydrocarbon products and no products which might have arisen from an isobenzofulvene have been isolated. Identification of the major photolysis product, recovered in ca 10% yield, is still in progress.

In view of a number of articles which report similarities between mass spectrometric and photolytic fragmentation pathways [29], it is interesting to note that although photolysis of the lactam 25 does not appear to generate 8,8­dimethyl isobenzofulvene, the major fragmentation in the mass spectrum (Figure 14) of 25 is a loss of HNCO.

Figure 14

3. Conclusion

It is apparent, from this work, that 2-heterocyclo[2.2.1]heptanes of the type 58 can readily be prepared by the cycloaddition of isobenzofulvenes to heterocyclic dienophiles. Such adducts, however, are prone to rearrangement to products of type 59, and that these compounds, in turn, often produce ring-opened products under mild hydrolytic conditions (Scheme 16). Notwithstanding, the cycloaddition route has proved to offer a simple route to indeno[1,2-c]furans and indeno[1,2-c]thiophenes and has a role in heterocyclic synthesis. Certainly, the availability of adamantyl derivatives using the recently described adamantylisobenzofulvene [31] should make products of this novel type appealing for biological testing programs.

While we were able to test the potential of only the 2-azabicyclo [2.2.1]heptanone 25 towards isobenzofulvene production, and that was unsuccessful, the other target molecules, the 2-thiabicyclo[2.2.1]heptanone 23 and the 2-oxabicyclo[2.2.1]heptanone 24 remain viable and present a synthetic challenge which is still to be realised.



Scheme 16

Accordingly, the value of molecules of the general type 58 as photochemical precursors to isobenzofulvene, must be severely limited.

4. Experimental