Bond Fixation in a [14]-Annulene: Synthesis, Characterization, and Ab Initio Computations of Furan Adducts of Dimethyldihydropyrene.

Reginald H. Mitchell,a* Yongsheng Chen,1a Vivekanantan S. Iyer,1a Danny Y. K. Lau,1a Kim K. Baldridge,1b and Jay S. Siegel1c*

Contribution from the Department of Chemistry University of Victoria, P. O. Box 3055, Victoria, British Columbia, the San Diego Supercomputer Center, P.O. Box 85608, San Diego, California, 92186-9784, and the Department of Chemistry, University of California, San Diego, La Jolla, California, 92093-0358.

Abstract: Furan adducts, 13, 15 and 16, of dimethyldihydropyrene are prepared in order to test the postulate that bicyclic annelations are generally effective at inducing bond localization in aromatic systems. A large 2 ppm down field shift of the NMR shifts of the internal methyl signals in 15 compared to 16 provides a significant indicator of bond localization in 15. X-ray diffraction analysis of 13 displays regular bond length alternation. Ab initio computations that do not include dynamic electron correlation are found to be inadequate for modeling the molecular structure of 1. Density Functional Theory models 1 well and predicts bond localization in derivatives of 1 consistent with the observed NMR spectroscopic and X-ray diffraction results.

 

Introduction

Bicyclic annelations (e.g. bicyclo[2.1.1]hexeno and oxanorbornadieneo) have been implicated as inducers of p-bond localization in aromatic systems. Dimethyldihydropyrene, 1, is an excellent test molecule for investigations of cyclic p-electron delocalized systems, because the NMR chemical shift of the internal methyl protons of 1, d -4.25, is relatively insensitive (< +/- 0.2 ppm) to simple substituent effects, but is very sensitive to anything that disturbs the cyclic delocalization of its p-electrons,. The methyl protons of 1 are shielded by 5.2 ppm from those in 2; this shielding is ascribed to the ring current caused by the cyclic delocalization of the 14 p-electrons in 1. Fusion of a strongly aromatic system such as benzene to give 3, causes substantial bond-fixation in the macrocyclic ring of 3, which betrays itself in a reduced ring current and reduced shielding of the methyl protons, d -1.63 instead of d -4.25. In contrast, the spectra of the cyclobutane annelated systems 4 and 5, with methyl signals clustered around -4 ppm, indicate that fusion of a nonconjugated carbocyclic ring causes no p-bond fixation; there is almost no change in methyl proton chemical shift between either 4 and 5 (d 0.12 ppm), or between these signals and those of 1 (d 0.16, 0.02 ppm), as would be expected if bond fixation were significant. Studies on a number of other nonconjugated carbocyclic annelated dihydropyrenes, all show insignificant effects on the ring current and thus presumably on the p-delocalization present (Figure 1).

 

Figure 1. Annelated dihydropyrenes 6-12 and their 1H chemical shifts (10-12 isomeric mixtures)

Results

Anomalous among the spectra of derivatives of 1 are the chemical shifts observed5 for the methyls of the two isomers of the oxanorbornadiene fused dihydropyrenes, 13 and 14; the signals for these methyls appear at d -3.34, -3.51 and -3.29, -3.45, respectively. A substantial reduction (16-17%) in ring current has occurred from that of 1. Isomer 13 was purified from the mixture by fractional crystallization, and a crystal structure was determined (full details are deposited). The relevant macrocyclic carbon-carbon bond distances are given in Table 1, together with the measured 3Jcis coupling constants, which are totally consistent with the X-ray data. Clearly, multiple techniques support substantial bond-fixation in the macrocyclic ring of 13, and by correlation of the NMR chemical shifts one can deduce the same to be true for 14.

Table 1. Selected C-C bond distance data and 3Jcis coupling constants for 13.

Bond Distance (Å) 3Jcis (Hz) Bond Distance (Å) 3Jcis (Hz)

1 - 2 1.365 8.42 8 - 13 1.428

2 - 3 1.406 6.97 13 - 14 1.359

3 - 4 1.357 14 - 15 1.427

4 - 5 1.413 15 - 16 1.374

5 - 6 1.357 8.79 16 - 17 1.413 6.93

6 - 7 1.417 17 - 18 1.376

7 - 8 1.361 18 - 1 1.409

A further test of whether ring current is actually reduced comes from the relationship between the chemical shift of the internal methyl protons and H-2 of 13. For a series of aromatic ring annelated examples, including 3, the chemical shift of the internal methyl protons is related to the chemical shift of H-2 through the equation:6

d(Me) = 17.52 - 2.69d(H-2)

In the 1H NMR spectrum of 13, H-2 appears at d 7.76, which by formula corresponds to d(Me) = -3.31, in extremely good agreement with the values found experimentally (-3.29 to -3.51). Such a correspondence among spectral shifts indicates that it is unlikely that the chemical shift of the methyls is affected by some anisotropic or geometric effect, rather than a ring current effect. That ring current effects are dominant is supported by comparison of the X-ray data for 13 with those of 1.6 Some of the relevant data are shown in Table 2.

Table 2. Bond angle data for part of 13 and 1

Angle 13 () 1 () Angle 13 () 1 ()

a 119.1 121.7 1-3-2 105.9 105.9

b 119.2 118.0 1-3-5 105.3 105.2

c 115.0 113.3 1-3-4 112.1 110.6

d 117.3 119.4

e 123.0 120.5

f 122.8 122.7

g 123.5 125.5

h 117.1 116.1

i 117.8 116.5

j 124.6 123.6

Clearly, the relationship between the methyl groups and the ring periphery in 13 is not very different from that in 1, in particular, the distance from the methyl group (atom 1 in Y) to the mid point of the 3-4 bond in Y is 1.98 A in 13 and 1. Thus, in both compounds, the methyl group is situated at approximately the same position relative to the center of the magnetic ring current field, and therefore any difference in chemical shift should be due primarily to a difference in ring current, and not a shift in position of the methyl group.

If indeed the oxanorbornadiene ring does have some special p-bond localizing effect, then synthesis and spectral characterization of the bis-adducts 15 and 16 should prove this effect. Comparison of the two resonance structures of 15A and 15B (symmetry nonequivalent contributors) and 16A and 16B (symmetry equivalent contributors), indicates that the p-electrons of 15 should manifest a smaller ring current effect than those of 16.

Reaction of the dibromide 17 with NaNH2 and t-BuOK in furan formed a 1:4 mixture of 15 and 16, three isomers of each (total yield 62%). In the 1H NMR spectrum of the mixture of 15 and 16, one set of internal methyl protons was centered around d -4, while the smaller set was around d -2.3, consistent with an enhancement of bond localization in one isomer type. Chromatography failed to separate the products, however, fractional crystallization several times from methanol-dichloromethane finally gave a sample with only the three isomers of 16 (Figure 2) and an enriched sample of the three isomers of 15. The gross structure of the isomeric mixture of 15 or 16 was easily proved by deoxygenation of the respective isomeric mixture using Fe2(CO)9 in benzene to the known dibenzannulenes. The three isomers of 16 could not be separated from each other, but the mixture gave a M+ at 364.148 (EI-HRMS), consistent with that required for C26H20O2 (364.146), and more convincingly this mixture gave only the known transoid dibenzannulene 18 on deoxygenation, thus proving the relative orientation of the annelations.

Figure 2. Isomers of 16 and the corresponding deoxygenation product 18.

The udud and the uudd isomers of 16 are symmetrical and each would be expected to display a single 1H NMR methyl resonance. The two methyl groups in the uduu isomer are however different, and thus two signals are expected for this isomer. Indeed, four methyl signals were observed, with the two signals of 16uduu (d -3.80 and -4.15) being of equal intensity, corresponding to 16uduu, the other signals (d -3.96 and -4.01) were not specifically assigned but belong to 16uudd and 16udud. The relative integrations of the three sets of signals was 5:1:2. After several recrystallizations, the ratio was 14:5:1. The 16uduu isomer is formed preferentially. Similarly, the three isomers of 15 also give four methyl signals. Analogously, 15uduu gave signals at d -2.17 and -2.48; the other two isomers gave signals at d -2.29 and -2.34. The relative integrations of these being 2:3:1; in this case the uduu isomer is not formed preferentially.

The NMR data indicate clearly that a substantial (35-40%) reduction of ring current has occurred for 15, and almost none for 16; 15B is predicted to be the dominant resonance contributor. Thus, the strong p-bond localizing effect of the oxanorbornadiene ring thus seems difficult to dispute.

Computational studies on 1 and derivatives are complicated by the size of the molecules and the difficulties associated with a correct computational description of the behavior of p-delocalized systems. For example, although several methods accurately predict the structure of benzene as D6h with 1.395 Å C-C bonds, few handle the thermochemistry with sufficient accuracy. Theoretical treatments of 1 are even more complex. Discarding semiempirical methods due to their unreliable nature, even restricted Hartree Fock (RHF) levels of theory at reasonable basis sets (6-31G(D) or DZ(2d,p)) predict a C2 (bond localized) structure for 1. These levels of theory describe the experimentally observed C2v structure as a transition state structure (one negative eigenvalue), ca. 10 kcal/mol higher in energy than the C2 form. Haddon predicted that electron correlation would be important for delocalized systems such as [10]-annulene. Indeed, only at dynamically correlated levels of theory such as Moller Plesset, MP2, or Density Functional Theory, DFT, levels of theory does the structure of 1, a [14]-annulene, come into accord with experiment, thus establishing self-consistency of theory for derivatives of 1. This level of theory is thus established as a minimum basis set necessary for meaningful comparison with experiment.

On the basis of annelation induced bond localization effects on benzene, systems 15 and 19 are expected to show roughly equal degrees of bond localization, as opposed to 20, which should show essentially no localization. DFT computations on 20, 15, and 19 substantiate this structural correlation unequivocally (Table 3). Whereas the bonds of the [14]-annulene in 20 all cluster around 1.40 Å and show no regular alternation in length, the bonds in 15 (and 19) group into two sets and alternate long-short around the annulene. Starting from the fused bond between the macrocycle and the annelation, one can highlight every other bond. As such, all highlighted bonds belong to the endo set, the others belong to the exo set. A parameter,

Dendo-exo =

can be defined in order to gauge the degree of bond localization. For 20, Dendo-exo equals ca. 1 pm and corresponds to localization of less than 5% of a bond. In contrast, for 15 and 19, dendo-exo is 3.3 and 3.1 pm respectively, corresponding to a localization of roughly 25% of a bond. These distortions are consistent with the kinds of distortions predicted and seen for 21 and 22.

Table 3. Selected Structural Parameters for 15, 19, and 20 from DFT computations.

Compound 1 20 19 15

dexo (Å) 1.411 1.402 1.391 1.389

dendo (Å) 1.411 1.412 1.429 1.428

Dendo-exo (pm) 0.0 1.0 3.8 3.9

dave (Å) 1.411 1.407 1.410 1.409

 

A perturbation analysis of the orbitals of cyclobutane interacting face-to-edge with the p-system of the macrocycle reveals a destabilizing filled-filled interaction between the s orbitals on the cyclobutyl fragment and the highest occupied molecular orbital (HOMO) of the macrocycle. Localization of the double bonds exo to the point of fusion reduces this destabilizing interaction. This explanation accounts well for the distortions in 22 and 20, but may be questioned for 15 and 21. Hydrogenation of the double bonds in the oxanorbornadieneo annelations of 13, 15 and 21 leads to 23, 24, and 25. Experimentally, the chemical shift perturbation vanishes for 23, and 24. Computationally, the bond localization is substantially reduced in 25, and by analogy we would expect the same for 23, and 24.. Therefore, 15/19 and 21/22 show similar structural distortions, but the components of their physical mechanism are potentially different. Although one can speculate that the anneleation effects in 19 act purely through s-p interactions and those in 15 act through a mixture of s-p filled-filled interaction and secondary through-space p-p and s-p interactions, a complete analysis of angle strain and orbital conjugation is needed. In 20, none of the correct orbitals are present at an energy which permits efficient interaction and no bond localization is seen.

Computational Details

The molecular structure of 1 has been determined at a variety of theoretical methods to determine self consistency. Reported here are the split valence 6-31G(d) and triple zeta valence TZV(d) basis sets, employed at the restricted Hartree-Fock (RHF) self-consistent field (SCF) level of theory. These basis sets include a set of six d polarization functions on all heavy atoms. These calculations were performed with the aid of the analytically determined gradients and search algorithms contained in GAMESS. Additional calculations at the MP2/6-31G(d) and Density Functional Theory levels were performed to determine the effects of dynamical correlation. The former method, a post-RHF method which incorporates correlation in terms of Moller-Plesset theory of order 2 (MP2), were performed using the GAUSSIAN 92 suite of programs. The Density Functional Theory (DFT) Methods, which inherently incorporate effects of correlation in their development, were performed with the aid of the numerical methods within Dmol. A double numerical basis set augmented by polarization functions, which is comparable in size to the 6-31G(d) basis set of the traditional Hartree-Fock methods, was chosen for the DFT calculations.

Experimental

The general experimental conditions are as previously described. The preparation of 13 has been reported.6

2,7-Dibromo-trans-10b,10c-dimethyl-10b,10c-dihydropyrene (17)

A solution of NBS (not recrystallized from water) (1.54 g, 8.62 mmol) in dry DMF (70 mL) was added slowly over 20 min to a stirred solution of dihydropyrene 1 (1.00 g, 4.31 mmol) in dry DMF (50 mL) at 0C and then the reaction mixture was stirred without further cooling for 2 h. Diethyl ether (200 mL) was added, and then the mixture was poured into ice-water. The ether extract was washed well with water, was dried (MgSO4) and silica gel was added to pre-absorb the product after evaporation. Chromatography on silica gel using ether-petroleum ether (1:10) gave the product as the major green fraction, 1.18 g (70%), which could be recrystallized from chloroform-methanol, mp 212-214C (lit.9 mp 213-214C). 1H NMR (360 MHz) 8.69 (s, 4), 8.51 (s, 4), -4.04 (s, 6); 13C NMR (90.6 MHz) 136.7, 126.8, 123.9, 118.5, 29.2, 14.2.

Furan adducts 15 and 16

NaNH2 (150 mg, 3.85 mmol) and t-BuOK (2 mg) were added to a stirred solution of dibromide 17 (122 mg, 0.31 mmol) and dried furan (4 mL) in dry THF (4 mL) at 20C under argon. After stirring for 48 h, methanol (0.5 mL) was added, and then silica gel (2 g) and then the solvent was evaporated. The solid residue was placed on the top of a silica gel column and chromatographed using first petroleum ether to elute any unchanged bromide and then ether-petroleum ether (7:3) to elute the products 15 and 16 , 70 mg (62%) as a mixture of six isomers. From the integrations of the internal methyl protons at -3.80 to -4.15 (16) and at -2.12 to -2.45 (15), the ratio of 16/15 was found to be 78:22 (4:1). Rechromatography of this mixture and fractional recrystallization several times from dichloromethane-methanol yielded 30 mg of pure 16 as a mixture of the three isomers shown in Figure 2. MS (CI) m/z 365 (MH+); EI-HRMS M=364.1485, C26H20O2 requires 364.1463. 1H NMR (360 MHz): 16uduu 8.46 (s, H-1,8), 8.43-7.96 (m, H-2,3,9,10), 7.23-7.16 (8 lines, H-6,13), 7.16-7.13 (8 lines, H-5,12), 6.59 (bs, H-4,11), 6.25 (bs, H-7,14), -3.80 and -4.15 (s, -CH3); 16udud + 16uudd 8.46 (s, H-1,8), 8.43-7.96 (m, H-2,3,9,10), 7.39-7.36 (6 lines, H-5,12), 7.34-7.31 (6 lines, H-6,13), 6.61 (bs, H-4,11), 6.23 (bs, H-7,14), -3.96 and -4.01 (s, -CH3) - see text. In the 13C NMR spectrum, each type of carbon showed three resonances, e.g. C-4: 83.97, 83.55, 83.50; C-7: 81.53, 81.31, 81.23, corresponding to the three isomers present. The cisoid isomers 15: 1H NMR (360 MHz, peaks after subtracting transoid isomers) 7.80 and 7.42 (s, H-1,4), 7.70-7.61 (m, H-2,3,9,10), 7.05-6.81 (m, H-6,7,12,13), 6.27 (bs, H-8,11), 5.92 (bs, H-5,14), -2.13, -2.27, -2.31, -2.45 (s, -CH3) - see text.

Deoxygenation of 16 to dibenzannulene 18

A mixture of mixed isomers of 16 (12 mg, 0.033 mmol) and Fe2(CO)9 (29 mg, 0.080 mmol) in benzene (4 mL, distilled from sodium, under argon) was stirred at 60C under Ar for 1 h. The cooled mixture was directly chromatographed on SiGel (5% water deactivated), quickly using petroleum ether as eluant. The deep blue band yielded 6 mg (55%) of 18, identical to an authentic sample10 (ms, nmr). When mixed isomers of 15/16 were used, both benzannulenes10 were obtained.

Hydrogenation of furan adduct 13 to 23

Recrystallized isomer 136 (40 mg) in ethyl acetate (10 mL) were added to pre-reduced Pt (5 mg) in ethyl acetate (10 mL), and the mixture was stirred under H2 for 30 min. Direct chromatography of the product on SiGel using petroleum ether as eluant gave 37 mg (93%) of product 23, as green crystals from petroleum ether mp 112-113 C; 1H NMR (300 MHz) 8.63 (d, 1H), 8.55-8.49 (m, 5H), 8.40 (s, 1H), 8.02 (t, 1H), 6.37 (m, 1H), 5.99 (m, 1H), 2.36-2.31 (m, 2H), 1.54-1.30 (m, 2H), -4.13, -4.23 (s, 3H each); 13C NMR (90.6 MHz) 142.2, 139.5, 139.3, 136.4, 136.1, 126.1, 124.0, 123.9, 123.3, 122.8, 122.5, 119.1, 114.4, 80.5, 78.1, 31.4, 30.3, 29.3, 27.9, 14.2, 13.7; UV (cyclohexane) lmax nm (e) 343 (60,000), 380 (23,200), 477 (4500), 640 (550); CI MS m/z 301 (MH+). Anal. calcd for C22H20O: C, 87.96; H, 6.71. Found: C, 87.30; H, 6.76.

Hydrogenation of mixed isomers 15/16

This was carried out exactly as described above for 13. The 1H NMR spectrum of the product containing 24 and the corresponding transoid isomer, indicated internal methyl protons at -3.94 and -4.36 with minor isomer peaks between -4.01 and -4.42, i.e. no peaks at higher chemical shift than -3.94. The other protons were as expected at 8.6-8.2, 6.3, 5.9, 2.4-1.3 as for 23.

Acknowledgments: Support was provided by the National Science Foundation (CHE9307582; ASC-9212619 and VPW to KKB) and the Canadian National Science and Engineering Research Council. A grant for supercomputing time was provided by the San Diego Supercomputer Center.

Supplementary Material Available: Tables and details of the X-ray analysis of 13 (15 pages). This material is contained in many libraries on microfiche, immediately follows this article in the microfilm version of the journal, can be ordered from the ACS, and can be downloaded from the Internet: see any current masthead page for ordering information and Internet access instructions.

Bond Fixation in a [14]-Annulene: Synthesis, Characterization, and Ab Initio Computations of Furan Adducts of Dimethyldihydropyrene.

Reginald H. Mitchell,1a* Yongshen Chen,1a Vivekanantan S. Iyer,1a Danny Y. K. Lau,1a Kim. K. Baldridge,1b and Jay S. Siegel1c*

Contribution from the Department of Chemistry University of Victoria, P. O. Box 3055, Victoria, British Columbia, the San Diego Supercomputer Center, P.O. Box 85608, San Diego, California, 92186-9784, and the Department of Chemistry, University of California, San Diego, La Jolla, California, 92093-0358.

[Supplementary Material- 15 pages]