Regiochemical Control in Triptycene Formation-An Exercise in Subtle Balancing Multiple Factors.

Reactions between 1,8-dichloroanthracenes with substituents in position 10 and ortho-chloroaryne afford mixtures of 1,8,13- (syn) and 1,8,16-trichlorotriptycenes (anti). The syn/anti ratio is dependent on these substituents. Electropositive substituents like SiMe3 and GeMe3 lead to preferred formation of the syn-isomer, whereas CMe3 groups exclusively afford the anti-isomer. Different quantum chemical calculations including location of transition states give conflicting results, but indicate the importance of dispersion forces for an at least qualitative prediction of results. The syn-trichlorotriptycenes with SiMe3 and GeMe3 substituents were characterized by using NMR spectroscopy, mass spectrometry, and X-ray diffraction experiments.

Triptycene represents one of a few rigid organic frameworks of D 3h symmetry without any (Lewis‐basic) heteroatoms. It was first synthesized by Bartlett et al. in 1942 using a multi‐step procedure starting from anthracene and p‐benzoquinone.1 In 1956, Wittig and Ludwig reported a more efficient access to triptycene in one step from anthracene by reacting it with in situ‐formed benzyne.2 The symmetry and rigidity of triptycene have inspired a plethora of applications in fundamental and applied chemical research.3, 4, 5 Substituted triptycenes are widely used, for example, as building blocks for fluorescent or non‐fluorescent organic macromolecules, polymers, and liquid crystals,3, 6 as rigid spacers in several Pd complexes used for cross coupling reactions,7 as devices in molecular machines,8 in crystal engineering processes,9, 10 and as a basis for the design of highly porous organic materials with numerous applications.11

Although the chemistry of triptycenes and their functionalization is generally in an advanced state, the 1,8,13‐trisubstitution motif remains a challenge for synthesis. However, exactly this pattern is interesting to introduce three functionalities oriented in the same direction. We try to make use of such 1,8,13‐trisubstituted triptycenes (also called syn‐triptycenes) as rigid organic frameworks for constructing directed polydentate Lewis acids,12, 13 but many other applications might be envisioned.

syn‐Triptycenes can be obtained through Diels–Alder reactions of 1,8‐disubstituted anthracenes with ortho‐functionalized arynes, a protocol introduced by Rogers and Averill in 1986.14 The drawback of this method is that the corresponding anti‐trisubstituted 1,8,16‐isomer is always formed as the main product when, for example, Cl‐functionalized anthracenes and arynes are used.12, 14 In 2010, we reported attempts to increase the syn/anti ratio by making use of the steric interference of the (bulky) anthracene substituent at C‐10 with the chlorine atom of the chloroaryne (Scheme 1). We expected this strategy to provide an increased formation of the syn‐isomer. However, the steric influence of the C‐10 substituent turned out to be minimal, whereas the electronic properties are dominant;12 of all substituents tested, the biggest R=C(CH3)3 led to the formation of 100 % anti‐isomer, despite the formation of an extremely deformed product by mutual repulsion of the Cl and R substituents, as indicated in Scheme 1 b.

Scheme 1

a) Proposed steric interactions of 10‐substituted 1,8‐dichloroanthracenes and chloroaryne; b) structural data for a tert‐butyl derivative.

Accepting that we cannot take steric control over this reaction, it is necessary to understand the electronic parameters determining it. Cycloadditions of substituted arynes have been studied experimentally and theoretically. Their reactions have been predicted by steric, charge‐controlled, and aryne‐distortion models; the latter two have recently been given preference and predict nucleophilic attack to be preferred at C‐3 in 3‐chlorobenzyne.15

First, simple DFT calculations [B3LYP/6‐31G(d,p)] indicated that the charge distribution of the carbon skeletons of the anthracene and benzyne can be manipulated by different substituents. However, the predicted partial charges depend very much on the quantum‐chemical method employed. Although in a recent theoretical study the charge distribution of chlorobenzyne has been given as +0.11 (C‐2) and −0.04 e (C‐3) [NBO charges, B3LYP/6–311+G(d,p)],15 we find contradictory results between natural bond orbital (NBO)16 (C2−0.02, C3+0.10 e) and atoms in molecules (AIM)17 (C‐2+0.01, C‐3−0.21 e) charges (B3LYP/6‐31G(d,p) [further details can be found in the Supporting Information, and electrostatic potentials (ESP) are also reported therein]. This demonstrates that the charge argument is less clear‐cut than suggested. More unambiguous are charges predicted for a range of 10‐substituted anthracene molecules: although the charge at C‐9 is virtually invariant for all molecules (R=SiMe3, CMe3, Ph, Cy, iPr, Me, H) (NBO −0.18 to −0.17, AIM +0.01 e for all), the charge at C‐10 is clearly more negative (NBO −0.44, AIM −0.65 e) for R=SiMe3 than for all carbon substituents (NBO +0.03 to +0.04, AIM 0 to −0.02 e). This distinction is clear, such that a SiMe3 group was likely to exert the desired regiochemical effect.

The limited number of available electropositive functions compatible with the other reactive groups in the systems restricts the choice of possibilities. However, 1,8‐dichloroanthracenes with EMe3 (E=Si, Ge) substituents in position 10 were accessible from earlier projects: 10‐bromo‐1,8‐dichloroanthracene can selectively be metalated with nBuLi and the carbanion reacted with Me3ECl reagents.18 The 10‐EMe3‐1,8‐dichloroanthracenes with E=Si (1) and Ge (2) were then reacted with in situ‐generated chloroaryne to obtain the corresponding trichlorotriptycenes (Scheme 2), which were then used to test the validity of the predictions.

Scheme 2

Conversion of the 10‐trimethylelement‐substituted 1,8‐dichloroanthracenes 1 and 2 with in situ‐generated chloroaryne to afford the trichlorotriptycenes 3 and 4. Reagents and conditions: 1) 3‐chloroanthranilic acid, isoamyl nitrite, DME, reflux, 4 h; 2) aqueous NaOH, MeOH, 55 % (3), 45 % (4). See the Supporting Information for details.

With respect to the earlier experiments with R=H, Me and CMe3 we observed a drastic increase in the syn/anti ratio of the triptycene product with R=SiMe3. Instead of 0:100 for R=CMe3, we now observe a strong preference for the syn‐isomer (for R=SiMe3 syn/anti mixture 84:16). A similar observation was made with R=GeMe3, but the preference for the syn‐isomer was less pronounced (syn/anti mixture 70:30). As both, Si and Ge, are more electropositive atoms than carbon, the original prediction by the calculations seemed to have proven true, and even the higher electronegativity of germanium compared to silicon seems to be reflected in the experimental results. Comparable results—a favored formation of syn‐Diels–Alder cycloaddition products—are observed when, for example, SiMe3‐ instead of CMe3‐substituted furane derivatives are converted with haloarynes.19

The isomers in the product mixtures of 3 and 4 could be separated by sublimation and were characterized by 1H, 13C{1H}, and 29Si{1H} NMR spectroscopy as well as high‐resolution mass spectrometry. In the case of anti‐3, the rotation about the CAr–Si axis is found to be hindered, as is indicated by two resonances at 0.99 (6 H) and 0.55 ppm (3 H), induced by the protons of the SiMe3 substituent. A similar splitting has been observed earlier for the tert‐butyl analogue of anti‐4.12 Characteristic for the two different isomers, syn and anti, are the singlet resonances of the bridgehead proton H9. That of syn‐isomer (syn‐4) experiences a larger downfield shift than that of the anti‐isomer (anti‐4) (δ=7.12 vs. 6.49 ppm).

The molecular structures of syn‐3 and syn‐4, as determined by single‐crystal X‐ray diffraction,20 are shown in Figure 1. Both compounds are isostructural. The molecules are of C 3 symmetry (close to C 3v) and exhibit paddlewheel configurations. Benzene and methyl substituents are arranged in a staggered conformation. The C−C bonds in the benzene rings vary by about 0.03 Å in length around that of benzene (1.395 Å21). Longer are the distances C(2)–C(3) and C(4)–C(5) [1.517(2), 1.548(2) Å (syn‐3); 1.519(2), 1.545(2) Å (syn‐4)]. The bonds C(4)–E(1) are significantly elongated compared to the corresponding standard C(sp3)–E distances [1.922(2) Å (syn‐3) vs. 1.87 Å and 1.994(3) Å (syn‐4) vs. 1.96 Å].22 This indicates intramolecular repulsion between the hydrogen atoms at C(6) and the methyl groups.

Figure 1

Molecular structures of syn‐3 and syn‐4 in the crystal. Side views (above) and views along the C(3)–C(4)–E(1) axis (below). Displacement ellipsoids are drawn at 50 % probability level. Hydrogen atoms are omitted for clarity. Selected bond lengths [Å] and angles [°] for syn‐3/syn‐4: C(1)–C(2) 1.386(2)/1.382(2), C(1)–C(8) 1.393(2)/1.399(3), C(1)–Cl(1) 1.744(1)/1.742(2), C(2)–C(3) 1.517(1)/1.519(2), C(2)–C(5) 1.408(1)/1.402(2), C(4)–C(5) 1.548(1)/1.545(2), C(4)–E(1) 1.922(2)/1.994(3), C(9)–E(1) 1.879(1)/1.954(2), Cl(1)–C(1)–C(2) 119.8(1)/120.0(1), Cl(1)–C(1)–C(8) 118.9(1)/118.9(1), C(1)–C(2)–C(3) 126.4(1)/126.6(2), C(2)–C(1)–C(8) 121.2(1)/121.0(2), C(2)–C(3)–C(2′) 105.6(1)/105.5(1), C(3)–C(2)–C(5) 113.8(1)/113.8(2), C(4)–C(5)–C(6) 127.2(1)/126.8(2), C(5)–C(4)–E(1) 114.2(1)/114.0(1).

The fact that electronic parameters dominate the regioselectivity of the reactions combined with the inability of charges to predict the reaction prompted us to gain a more detailed view of this aryne [4+2] cycloaddition reaction. The most direct way of theoretical investigation of reactions and their product distribution is the calculation of structures and energies of the corresponding transition states. We performed such calculations for a series of reactions by using different approximations. To establish a benchmark and to find the most suitable level of theory, we calculated the barriers of activation for the simplest system first: the reaction of 1,8‐dichloroanthracene with chlorobenzyne (for details, see Table S4). It turned out that the best and still affordable method is DFT using the B3LYP functional with the 6‐31G(d,p) basis set including corrections for basis‐set superposition error. The optimized transition‐state structures are shown in Figure 2.

Figure 2

Optimized [B3LYP/6‐31G(d,p)] transition‐state structures for the formation reaction of 1,8,13‐trichlorotriptycene (syn, above) and 1,8,16‐trichlorotriptycene (anti, below).

AIM and interacting quantum atoms (IQA) techniques23 revealed interesting aspects for these transition‐state structures (Figures S4 and S5). Bond critical points (BCP) appear during bond formation in the reactions, with their properties being those of weakly stabilizing closed‐shell interactions. Natural energy decomposition analysis (NEDA) in NBO theory24 shows better stabilization for the transition states of anti‐trichlorotriptycene than for the syn‐isomer (Table S6). The lower activation barrier makes anti‐trichlorotriptycene the preferred product in the corresponding reaction. Detailed analysis reveals the importance of the electronic component (sum of electrostatic, polarization, and self‐energies) in this better stabilization. Relating the relative energies for the transition states for syn‐ and anti‐isomers to the corresponding experimentally observed syn/anti product ratios in reactions of chlorobenzyne with 1,8‐dichloroanthracene (Table 1), the B3LYP/6‐31G(d,p) results so far supported at least qualitatively the observations.

Table 1

Results of the trichlorotriptycene syntheses by conversion of the corresponding 1,8‐dichloroanthracene derivatives 1 and 2 with in situ‐generated chloroaryne. The data for other 10‐substituted 1,8‐dichloroanthracenes (R‐10=CMe3, Me, H) are given for comparison.12 The reactions have been also investigated by quantum‐chemical calculations (see the main text and the Supporting Information). All yields are given for isolated mixtures of syn‐ and anti‐trichlorotriptycenes.

Compound	R‐10	Syn [%]	Anti [%]	Yield [%]	Product
	H	21	79	16
	Me	37	63	42
	CMe3	0	100	43
1	SiMe3	84	16	55	3
2	GeMe3	70	30	45	4

However, we could not achieve even a qualitatively correct prediction of the product ratio for the reaction with R=SiMe3 (Scheme 2). Consequently, a substantial series of additional calculations was performed (Table S7) to find more reliable transition‐state energies for this reaction. All attempts of expanding the basis set, using different DFT functionals and modeling in solution, did not improve the calculated energies: the barriers to the formation of the syn‐isomer (with R=SiMe3) were still predicted higher than those for the corresponding anti‐isomer. Attempts to account for static correlation with the CASSCF method did either not give results compatible with the experimental findings. However, single‐point MP225 and XMCQDPT2 energies26 for the transition‐state structures from the respective RHF and CASSCF calculations indicated that dynamic correlation plays a significant role in these reactions. To test this hypothesis further, we carried out very time‐consuming MP2/def2‐SV(P) optimizations for the transition states. These revealed that the transition states for the formation of syn‐ and anti‐trichloro‐10‐(trimethylsilyl)triptycene (3) can have completely different structures (Figure 3: TS to the syn‐isomer) to those predicted by DFT calculations. The validity of this transition‐state structure for the syn‐isomer on this particular level of theory has been verified by computing the full reaction path.

Figure 3

Different views of the optimized [MP2/def2‐SV(P)] transition‐state structure of the formation of syn‐1,8,13‐trichloro‐10‐(trimethylsilyl)triptycene (syn‐3).

Figure 3 shows the structure of the transition state of the reaction of 1 with chloroaryne to syn‐1,8,13‐trichloro‐10‐(tri‐ methylsilyl)triptycene (3). Surprisingly, we see that it is likely stabilized by π‐stacking of the benzyne ring with the anthracene molecule. Unfortunately, all our attempts to find computationally and to fully prove the existence of a transition state to the anti‐isomer of 3 were not successful.

Despite its seeming simplicity, the formation of triptycenes from the reaction of arynes with anthracenes turns out to be a highly complex system governed by many parameters that require a subtle balance. The finding of a distinct contribution of dispersion in the transition state sheds new light on such reactions. We will now examine further tailor‐made reference systems experimentally and theoretically in order to gain an increasingly valid description of the multiple factors that determine the regioselectivity of such reactions.

Conflict of interest

The authors declare no conflict of interest.

As a service to our authors and readers, this journal provides supporting information supplied by the authors. Such materials are peer reviewed and may be re‐organized for online delivery, but are not copy‐edited or typeset. Technical support issues arising from supporting information (other than missing files) should be addressed to the authors.

Supplementary

Click here for additional data file.