Anthony G.
Avent
,
Paul R.
Birkett
*,
Adam D.
Darwish
,
Harold W.
Kroto
,
Roger
Taylor
* and
David R. M.
Walton
School of Chemistry, Physics and Environmental Sciences, Sussex University, Brighton, UK BN1 9QJ
First published on 23rd November 2000
Reaction of a CCl4 solution of C70Ph8 with air in the presence of light and [70]fullerene gives a bis-lactone C70Ph8O4 (44%), together with a 1,2-diol, C70Ph8(OH)2, 19,26,33,37,45,49,53,63-octaphenyl-7,8,19,26,33,37,45,49,53,63-decahydro[70]fullerene-7,8-diol (40%), 1 of Cs symmetry. This latter is also produced from reaction of a dichloromethane solution of C70Ph8 with 18-crown-6 and KMnO4, together with another Cs symmetry diol, 2, two possible structures for which are proposed; both 1 and 2 are the first diols to be derived from [70]fullerene. Whereas 1 is very insoluble in most organic solvents (but dissolves in THF), 2 is readily soluble in many organic solvents such as toluene, benzene, chloroform, carbon disulfide, and THF. The derivatives show appreciable differences in both their NMR spectra and their EI mass spectrometry fragmentation patterns.
On standing, C70Ph8 (produced by electrophilic substitution of C70Cl10 into benzene, with concurrent Cl2 loss) undergoes spontaneous oxidative cage-opening to give a bis-lactone in a reaction which may involve the intermediate formation of a bis-vinyl ether.5 During the preparation of C70Ph10 and C70Ph8 from C70Cl10, we isolated C70Ph9OH, the first fullerene to be described bearing a single hydroxy group attached to the cage.6 Spontaneous oxidation of C60Ph5H produces a benzo[b]furano[60]fullerene,7 and now we report that spontaneous oxidation of C70Ph8 produces in addition to the bis-lactone described above, a 1,2-diol, C70Ph8(OH)2.
Hirsch and co-workers have also reported the formation of a 1,2-diol from a cyclopropanated [60]fullerene derivative.12 We can also make diol 1 by their method, together with a second diol 2.
The diol 1 can also be made by the crown ether oxidation procedure described by Hirsch and co-workers.12 18-Crown-6 (3.95 mg, 1.24 equiv.) and a 9.3 × 10−3 mol solution of potassium permanganate (1.61 cm3, 1.5 × 10−5 mol) were added to C70Ph8 (17.6 mg, 1.21 × 10−5 mol) dissolved in dichloromethane (20 cm3) and the mixture was stirred vigorously. The colour changed to light yellow, and the mixture was then treated with acetic acid (2 cm3) and stirred for 30 min. The organic layer was separated, washed with water (10 cm3), and dried (MgSO4). The yellow solid remaining after removal of the solvent was dissolved in carbon tetrachloride, applied to a silica-gel column, and eluted with this solvent to give first a small quantity of unreactive impurity present in the starting material. Elution first with dichloromethane gave a mixture of two products (set aside for separation), and then with dichloromethane–methanol (94∶4) gave 1. The former mixture was re-purified by further chromatography (dichloromethane elution) and gave diol 2 as the main component (17% overall, Rf = 0.79); another component was present in very low yield and was not characterised.
![]() | ||
Fig. 1 EI mass spectra (70 eV) for (a) compound 1 and (b) compound 2. |
![]() | ||
Fig. 2 IR spectra (KBr) for (a) compound 1 and (b) compound 2. |
![]() | ||
Fig. 3 1H NMR spectra for (a) compound 1 and (b) compound 2. |
NOE analysis showed that the four sets of ortho hydrogens at δ 7.97, 7.63, 7.52, and 7.46 are due to phenyl rings D, C, A, and B, respectively (Fig. 4); the meta and para hydrogens were identified by 2D-COSY (both Double Quantum Filtered, and Hartmann–Hahn). As with other phenylated fullerenes (see e.g. ref. 4), the more downfield the signal for the ortho hydrogens for a given ring, the more downfield are the corresponding signals for the meta hydrogens relative to those for the para hydrogens. Thus here the differences δm − δp are ca. 0, 0.45, 0.8 and 0.9 for rings D, C, A and B, respectively. For C70Ph9OH, the signals for the phenyl groups adjacent to the hydroxy group (and which were thus unambiguously identified) were notably downfield relative to those for the other phenyl groups. Likewise in 1, the resonances for the phenyl groups (D) adjacent to the electron-withdrawing hydroxy groups are the most downfield (by ca. 0.17 ppm compared to the corresponding peaks in C70Ph8).3 However, the (downfield→upfield) sequences of positions in the NMR spectrum for the ortho hydrogens of the corresponding phenyl groups A–D differ for the C70Ph8, C70Ph10, C70Ph9OH and 1, being DBCA, BACD, DABC, and DCAB, respectively. Clearly there is much yet to be learned concerning the operation of electronic effects in fullerenes.
![]() | ||
Fig. 4 Schlegel diagram for diol 1 showing the NOE couplings. |
![]() | ||
Fig. 5 Schlegel diagram for diol 2 showing the NOE couplings. |
Comparison between the two isomers of the resonance positions for the ortho hydrogens of equivalent rings shows that the resonances for ring A are in almost the same position in both isomers, consistent with ring A being most remote from the OH groups. Likewise the resonance for ring D is marginally more downfield in isomer 1 since this ring is nearer to the presumed location of the OH groups. Surprisingly, however, the resonances for ring C are more downfield in isomer 1, whilst those for ring B are more upfield; there is no obvious explanation for this.
The signals for the sp2 carbons on the symmetry plane at 176.86, 145.35, 144.65 and 129.10 ppm are widely separated and much more so than for either C70Ph8 or C70Ph10. It is reasonable to assume that the 1,2-bond (see Fig. 4) is the one that is highly polarised, the resonance at 176.86 ppm being due to C1 and that at 129.10 being due to C2. No comparable analysis was possible for C70Ph9OH6 because the C1 symmetry makes all of the sp2 carbon resonances indistinguishable.
Our assignment of structure for isomer 1 is based upon: (i) its formation during the spontaneous oxidation that produced also the cage-opened bis-lactone, which involved oxidative cleavage of the 7,8-bond; (ii) NOE evidence that the OH groups are in close proximity to the phenyl groups D; (iii) the similarity of the structure to that (unambiguously determined) for C70Ph9OH,6 differing only in replacement of one phenyl group by OH. We therefore assign isomer 1 as 19,26,33,37,45,49,53,63-octaphenyl-7,8,19,26,33,37,45,49,53,63-decahydro[70]fullerene-7,8-diol.
1. Diol 1 is much less soluble than diol 2. This suggests that there is a significant difference in effective molecular size and hence hydrogen bonding between the two diols. It is possible to conjecture various scenarios involving either increased, or decreased intermolecular hydrogen bonding due to structural differences. Consequently this approach is not fruitful. We can conclude that there must be a significant structural difference between the two isomers, either in the relative locations of the hydroxy groups, or in the proximity of the phenyl groups to the hydroxy groups. We re-address these points below.
2. In general, diol 1 exhibits overall more downfield features in the NMR spectra compared to diol 2, and this is particularly marked for the most downfield resonance in each spectrum. Thus for 1 and 2, respectively, these are 7.97, 7.90 (ortho-H, ring D), 140.13, 139.49 (ipso-C), 95.18, 80.66 (sp3-COH), 176.86, 157.42 (on-plane sp2-C). The locations of the OH groups for diol 1 (shown in Fig. 4) are the most consistent with these data. In particular, the OH groups are nearest to the phenyl groups D, nearest to one of the on-axis carbons, and adjacent to each other causing mutual downfield shift of the resonances for both the OH (probably aided by hydrogen bonding) and for the (sp3-C
OH) groups.
3. Two possible structures are then available for diol 2. One is as shown in Fig. 5, the other in Fig. 6. The structure in Fig. 5 involves 1,2-shifts of the OH groups (or more probably, the oxygens in the precursor). In this the OH groups are no longer adjacent, so that different physical properties could be expected. The OH groups lie midway between the on-plane sp2 carbons, and so could be expected to cause the least downfield shift of their resonances. Also, the OH groups would not mutually withdraw electrons from each other, so that the sp3-COH carbons should appear more upfield as observed. The OH groups are sufficiently near to the hydrogens of phenyl groups D to account for the observed NOE effect.
![]() | ||
Fig. 6 An alternative structure for diol 2. (NB: In order to keep the numbering consistent for the three diol structures, lowest locant numbering is not used here.) |
In the alternative shown in Fig. 6, the OH groups are in a 1,3-relationship compared to those in diol 1. They could therefore arise from normal 1,3-rearrangements, or, since the OH groups are adjacent, from the usual alkene–KMnO4→diol mechanism. Since the OH groups are adjacent as they are in 1, the sp3 resonances for the carbons to which they are attached should probably appear well downfield as for isomer 1, with similar solubility being observed (which is not the case). Moreover, they would also be in the same relationship to one of the on-plane sp2 carbons as for isomer 1, so the resonance for this carbon should presumably also appear much more downfield. (These arguments neglect the effect of the phenyl groups which are nearer to each of the carbons described above.) Since the distance between the OH groups and the ortho-hydrogens of phenyl ring D is significantly greater than in the above alternative, the observed NOE enhancement may be harder to rationalise; however, the OH group may be pointing towards phenyl group D so that an interaction may nevertheless be feasible.
Overall, there is no overwhelming evidence favouring either of the two possibilities, and no clear decision can be made between them.
We considered the possibility that in view of the markedly different mass spectrum for ‘diol’ 2, and its greater solubility, it might be a dihydro compound. This can, however, be ruled out on four grounds. (i) Whilst we have detected many dihydro derivatives from arylation of fullerenes involving halogeno precursors, this is readily rationalised since halogenation is a radical process. Hence by the principle of microscopic reversibility, dehalogenation must also be a radical process, leading to fullerene radical intermediates which will abstract hydrogen from their surroundings in the customary radical manner. However, our oxidation does not involve any halogenofullerenes. (ii) The preparation involves oxidising conditions, hence formation of hydrides would not be expected. (iii) Both the 13C and 1H NMR resonances for C-OH (80.66 and 6.48 ppm, respectively) are inconsistent with those for known hydrides. For example, in C60Ph5H the corresponding values are 58.3 and 5.2 ppm, yet [60]fullerene is considerably more electron-withdrawing than [70]fullerene (as shown by the difference in chemical shifts for the di- and tetra-hydrides).14 (iv) We have obtained the mass spectra for a wide range of dihydrofullerenes and these show an invariant feature, viz. a parent ion at M amu and a major fragment at M − 2 amu. No such feature is apparent in Fig. 1a.
This journal is © The Royal Society of Chemistry 2001 |