Syntheses of bicyclo[3.3.0]octanes and bicyclo[4.3.0]nonanes by ring expansion of isopropylidenecyclobutanes

When subjected to HBr/HOAc in polar solvents like acetic acid, 6-(1-methylethylidene)- bicyclo[3.2.0]heptanes undergo a ring expansion reaction yielding 2-bromo-3,3-dimethylbicyclo[3.3.0]octane and 3-bromo-2,2-dimethylbicyclo[3.3.0]octane. Several other isopropylidenecyclobutanes have been found to undergo the same reaction with high stereoselectivity and moderate regioselectivity. In less polar solvents like diethyl ether the ring expansion reaction is suppressed, and bromides resulting from addition of HBr to the isopropylidene double bond are obtained.


Introduction
The bicyclo [3.3.0]octane and bicyclo[4.3.0]nonane][10][11][12] Despite several existing methods, the structural variety of these compounds still calls for new practical procedures to be developed. 13While working on a synthesis of the insect pheromone component lineatin, we found that the epoxide of 1 gave an acid catalysed ring expansion. 14Later we found that using HBr in acetic acid gave a near quantitative yield of the ring expanded product 2. 15 The reaction was found to be both stereo-and regioselective as seen from both spectroscopic data and X-ray crystallography.Previous attempts in our group to achieve ring expansion of compound 1 using protic acids like HCl, p-toluenesulfonic acid or CF 3 COOH, and Lewis acids like BF 3 , AlCl 3 , HgSO 4 , Hg(OAc) 2 or AgNO 3 were unsuccessful. 15However, using 45 % HBr in acetic acid a near quantitative yield of a product corresponding to compound 5 was achieved.When the reaction was carried out with 33 % HBr in acetic acid at room temperature using the same amount of HBr (~8 eq.), a mixture of products were obtained.The reactions were finished in 0.5-2 h and three products were observed.Two of these were ring expanded compounds 5 and 6.In addition variable amounts of 7 resulting from addition of HBr across the double bond, were also seen (Scheme 4).It was observed that 7 rearranged on the GLC, and for this reason it was not possible to give exact amounts of these compounds.The 1 H NMR spectrum of the product mixture resulting when the alkene 4d was used as the substrate, indicated that the ratio of the ring expanded compounds (5d + 6d) to 7d was approximately 70:30, and that the ratio of 5d to 6d was 58:42 ( 1 H NMR). Prolonged reaction times did not change the ratio 5d:6d.When substrate 4e was used, the ratio of the ring expanded compounds (5e+6e) to 7e was approximately 90:10.The gem-dimethyl singlets are easily detectable in the 1 H NMR spectra of the product 7.So when none of these resonances were seen in the spectrum of the reaction mixture using 4a as substrate, this was clearly indicating that none or only small amounts of 7a could have been formed.
Attempts to isolate 5a and 6a by column chromatography failed since no separation was achieved, and separation of 5 and 6 by chromatography was not attempted.Instead analytical samples of 5 and 6 were isolated using preparative GLC.
Generally a high stereoselectivity was achieved.According to both 1 H NMR and GLC analyses mainly one stereoisomer was formed, and only a few per cent of the other isomer could be detected.Representative examples are depicted in Table 2.The compounds 5, 6 and 7 are easily identified from their respective 1 H NMR spectra.The 1 H NMR spectra of 5 exhibited a characteristic doublet at δ 3-4 ppm due to the CH-Br signal.In the spectra of 6 the corresponding signal appeared as a doublet of doublet at δ 3.8-4.5 ppm.The compounds 7 could be identified from the two methyl singlets at δ 1.6-1.7 ppm consistent with a gem-dimethyl group situated on the same carbon atom as the bromine atom.The other features of the spectra were also in accord with the structures.
The rearrangement gave mainly one stereoisomer, but due to the flexibility of the two fused 5-membered rings it was not possible to use coupling constants to confirm which stereoisomer was predominantly formed.However, thorough analysis of the NMR spectra of 5a made it possible to distinguish the two protons on C4.A fairly strong interaction between the endo H4 proton and the α-proton (H2) based on the ROESY spectra could be seen, tentatively showing the stereochemistry of the bromine substituted carbon atom (H2) as depicted in Figure 1.The regioselectivity, however, was only moderate and best for the isopropylidenecyclobutane 4e, assumed to be the most strained substrate.The least strained substrate 4d yielded the lowest selectivity.With the substrates 4a and 4e only minor amounts (<10 %, GLC) of side products were observed.With the substrate 4d up to 18% side products were present (GLC), but some of them may result from decomposition of 7d in the injector.The substrates 4b and 4c, however, gave mixtures of several unidentified products where the ring expanded products 5 and 6, according to GLC analyses, constituted only small amounts.This was probably due to addition of HBr to the endocyclic double bond.Small amounts of two unidentified compounds could be isolated by preparative GLC from the complex mixture resulting from substrate 4b.The 1 H NMR spectra indicated that no double bonds were present in these compounds, and no attempts were made to further elucidate the structures.The reaction mixture resulting from substrate 4c was so complex that separation was not attempted.
Changing the temperature of the reaction resulted in only minor effects.(Table 3) Both the stereo-and regioselectivity of the reaction was the same as at room temperature.Temperatures ranging from 0-5 °C to 70 °C were tried.For substrate 4d (entry 7), however, lowering the temperature to 0-5 °C slowed the ring expansion reaction down, and the major product was 7d (GLC) where the ring expansion had not taken place.The amounts of side products formed were approximately the same as at room temperature.Unfortunately, lowering the temperature did not affect the outcome of the reaction for the substrate 4c (entry 5), and a complex mixture containing only minor amounts of 5c and 6c resulted.Elevation of the temperature (entry 4) gave no trace of 5c and 6c.Since the temperature effects were minimal, changing the polarity of the reaction medium was tried.Representative results are presented in Table 4.
At first the reaction was performed using the same amount of HBr (in acetic acid) as before (~8 eq.).Using substrate 4a as a model, solvents with polarities ranging from hexane to CH 2 Cl 2 were added in a ratio of HBr/HOAc : solvent, ~1:3 (eg.entries 1 and 2 ).The regioselectivity did not improve.Moreover, using diethyl ether as the solvent, the ring expansion reaction was suppressed completely yielding 7a as the only product identified.Only minor amounts of side products (<10%) were observed.The reactions were performed at room temperature except for entry 6 (substrate 4c) that was performed in refluxing ether.Comparison of GLC chromatograms of the reactions of the bromide 4c at room temperature and at reflux, indicated that the temperature change only resulted in minor differences in the product ratio.Purification of 7a by preparative GLC or flash chromatography failed, and only the ring expanded products 5a and 6a were isolated.Even at direct injection on the MS, rearrangement of 7a was observed.The compound 7b gave a spectrum that was tentatively associated with the structure depicted for this compound, but for 7c and 7d no attempts to measure MS spectra were made since they all rearranged as easily as 7a.When the reaction was performed in diethyl ether using an excess of only 2-4 eq. of HBr (HBr/HOAc:ether, ~1:20) (entries 4 to 7) no change in the outcome of the reaction was observed; the ring expansion reaction was suppressed for all the substrates, and only 7 were obtained.No attempts were made to purify 7b-d since the purification of 7a failed.The compound 7c was not isolated, but merely identified from the 1 H NMR spectrum of the crude product by resonances at δ ~1.6-1.7 ppm corresponding to the gem-dimethyl group situated on the bromine substituted carbon atom, a singlet at δ 1.85 ppm corresponding to the vinylic methyl group and a multiplet at 5.27-5.37ppm (alkene proton).Signals due to formation of the rearranged bromides 5c and 6c could not be seen in the spectrum.The yields of the products 7a-7d have not been optimized.
Slower addition of the HBr/HOAc solution resulted only in a slower reaction, and in accordance with literature, 23 an excess of 2-3 eq. of HBr was necessary to complete the reaction.
The stereochemistry of the bromides 7 was difficult to establish, but the ROESY spectrum of 7b shows a strong coupling between the two bridgehead protons H1 and H5, and a weaker coupling between the bridgehead proton H5 and the α-proton (H6).Molecular models (ball-andstick models) indicate that due to the rigidity of this bicyclic compound, the coupling between protons H5 and H1 and between protons H5 and H6 should be of similar strength if the α-proton (H6) and the bridgehead protons are syn.This indicates that 7b has the stereochemistry depicted in Figure 2 with the (CH 3 ) 2 CBr-group situated exo.This is confirmed by the ROESY spectrum revealing correlations between the (CH 3 ) 2 CBr-group and both the bridgehead proton H5 and the exo H7 proton.

Figure 2
Finally, attempts to achieve ring expansion on 7b and 7c were made treating them with acetic acid at elevated temperatures.The substrate 7b yielded the ring expanded compounds 5b and 6b in moderate regioselectivity.The substrate 7c gave a complex mixture containing moderate amounts of 5c and 6c (Table 5 and Scheme 5  The reaction gave an impure mixture, and the 1 H NMR spectrum of this was too complex to indicate the conversion of 7c or the ratio of 5c and 6c formed.On the other hand, the crude mixture obtained from 7b gave consisting results, when analysed by GLC and NMR, both with respect to conversion of the starting material and the ratio of 5b to 6b.This information is indicative of both the conversion of 7c and the ratio of 5c and 6c, although the bromide 7c has been found to rearrange on the GLC. Preparative GLC yielded analytical samples of 5b, 6b and 5c.For 6c an impure sample was obtained, and 6c was merely identified from the 1 H NMR spectrum of this sample by the singlets at δ 0.96 and 1.16 ppm (the gem-dimethyl groups), a multiplet at δ 1.69-1.77ppm (alkene CH 3 group), a doublet of doublet at δ 4.11 ppm (CHBr proton, J 5.4 and 5.9 Hz) and a multiplet at δ 5.25-5.35ppm (alkene proton).
A possible mechanism of the ring expansion reaction is depicted in Scheme 6.The initially formed tertiary carbocation can rearrange through either pathway a or b yielding 5 or 6, respectively.This mechanism fails to explain the high stereoselectivity exhibited by the reaction, however.Sterical congestion alone cannot explain the high stereoselectivity, and possibly a cage type mechanism is at work.
When the reaction was performed with the isopropylidenecyclobutane 1, a higher regioselectivity was reported. 15This may be due to the fact that if substrate 1 were to undergo a ring expansion reaction by pathway b, a severely sterically congested bromide with adjacent gem-dimethyl substituted carbon atoms would result.However, the mechanism of the reaction was not studied.

Scheme 6 .
Scheme 6. Possible mechanism of the ring expansion reaction.

Table 1 .
Starting materials

Table 2 .
Treatment of the isopropylidenecyclobutanes with excess 33 % HBr in acetic acid at room temperature a Conversion 100 %.Ratio based on GLC analyses (at full reaction time) and 1 H NMR data (of the crude mixture).

Table 3 .
Temperature effects a Conversion 100 %.Ratio based on GLC data.b i. e. <15% c Rearranges to a certain extent on the GLC.

Table 4 .
Solvent effects d Slow addition of HBr in acetic acid; e Reaction performed at reflux

Table 5 .
Ring expansion of HBr adducts a Conversion based on GLC data.b Conversion based on 1 H NMR data.