2,7,11,16‐Tetra‐tert‐Butyl Tetraindenopyrene Revisited by an “Inverse” Synthetic Approach

Abstract

A new synthetic route to tetraindenopyrene (TIP)—a bowl‐shaped cut‐out structure of C₇₀—is reported. The key step in this approach is a fourfold palladium‐catalyzed C−H activation that increases the yield more than 50 times in comparison to the approach originally described by Scott and co‐workers. Besides examination of its optoelectronic properties and study of its aggregation in solution, TIP was also re‐investigated by dispersion‐corrected DFT methods, which showed that dispersion interactions significantly increase the bowl‐to‐bowl inversion barrier. Furthermore, TIP was used as a semiconductor in p‐channel thin‐film transistors (TFTs).

Bowled over: An “inverse” route to a tetraindenopyrene (TIP), which is a cut‐out structure of C₇₀, is presented. The compound was reinvestigated by DFT calculations and tested as a semiconductor in p‐channel organic thin‐film transistors.

The discovery of fullerene C₆₀ in 19851 stimulated chemists to synthesize this molecule,2 other fullerenes or cut‐outs thereof—the so called buckybowls.3 Buckybowls have interesting properties themselves, but have also been used for the bottom‐up synthesis of fullerenes, for example, by flash vacuum pyrolysis.4 Besides corannulene,5 sumanene6 is the simplest substructure of C₆₀ and therefore it is not surprising that these two compounds are the most frequently studied.3a In contrast to C₆₀‐related buckybowls, similar approaches to compounds representing substructures of C₇₀ (Figure 1) are much rarer.7 In this respect, Kuo's C₃₈H₁₄ and C₄₀H₁₄ bowls are among the largest realized so far, with bowl depths of up to 2.33 Å.8 Other substructures of C₇₀ 9 such as rubicenes10 or dibenzorubicenes show smaller bowl depths of 1.68 Å10b or twisted10a, 11 conformations in the solid state. Some of these molecules have been employed as the semiconductor in p‐channel thin‐film field‐effect transistors, with a maximum reported charge‐carrier mobility of 1 cm² V⁻¹ s⁻¹).11

Figure 1.

Top left: structure of C₇₀ with tetraindenopyrene (TIP) 2 highlighted in red. Right: synthetic approach by Scott,13 bottom: this work.

Another substructure of C₇₀ is tetraindenopyrene (TIP, 2, Figure 1), which had been the subject of theoretical investigations by Havenith et al.12 and was later synthesized by Scott and co‐workers.13 The key step of Scott's TIP synthesis was a quadruple Pd‐catalyzed direct arylation of pyrene 1, which provides a yield of only about 0.5 % of TIP 2. In the same publication, a one‐pot procedure from 1,3,6,8‐tetrabromopyrene and 2‐bromophenyl boronic acid was described, but yields were again in the range of 0.5 %.13 Despite the very low yield of the cyclization step, the photophysics of TIP 2 were thoroughly investigated at that time. Based on these properties, the authors suggested that TIP may be a potential candidate for organic electronics or materials chemistry, such as long wavelength dyes for special high‐temperature applications. Very recently, the synthesis of a structurally related tetra‐n‐octyl TIP by an alumina‐mediated HF elimination was reported.14 Unfortunately, neither full characterization nor detailed discussion on photophysical or electrochemical properties were included. It is known that the K region of pyrene has a substantial olefin character, and thus the C−H activation may occur by a Heck coupling mechanism, rather than by a C−H activation in which the hydrogen is abstracted from a benzene ring.15 Therefore, we developed an alternative approach towards TIP 2 based on C−H activation of tetrachloropyrene 3 (Figure 1).16

The synthesis of pyrene derivative 3 was described previously, starting from pyrene in six consecutive steps.16, 17 We developed a different synthetic route for 3, starting from the commercially available hexahydropyrene 4 (Scheme 1) which was selectively fourfold‐brominated to 5 and isolated in 84 % yield by simple filtration.18 Subsequent Suzuki–Miyaura cross‐coupling under Fu conditions (Pd₂dba₃, HPtBu₃BF₄) and oxidation of the unsaturated propylene tethers by DDQ gave pyrene 7 in 56 % yield over two steps (for details, see the Supporting Information).

Scheme 1.

Synthesis of tetrachloro tetraaryl pyrene 3. a) Br₂, Fe, CH₂Cl₂, 80 °C, 30 min; b) 4‐tBuPhB(OH)₂, 8 mol % Pd₂dba₃, 30 mol % HPtBu₃BF₄, THF, K₂CO_{3 aq} (1 m), 80 °C, 16 h; c) 3 equiv DDQ, toluene, 130 °C, 4 h; d) NCS, CHCl₃, 80 °C, 42 h.

The next step was the tetrachlorination of pyrene 7, which was achieved with a slight excess (4.5 equiv) of N‐chlorosuccinimide (NCS) in chloroform to obtain 3 in 94 % yield (Scheme 1). It is worth mentioning that the chlorination was described previously using a large excess (>50 equiv) of sulfurylchloride.16 With the modified synthetic route described here, pyrene derivative 3 can be synthesized in just four steps from commercially available hexahydropyrene 4 in an overall yield of 44 %. Furthermore, no purification by column chromatography is required, allowing the synthesis of 2 on gram scale. In comparison, the previously described synthetic route started from pyrene with an overall yield of 4 % in six steps and required two steps of purification by column chromatography.16, 17, 19

All compounds were fully characterized (see the Supporting Information). By vapor diffusion of n‐pentane into saturated solutions of 5 and 6 in dichloromethane, crystals of sufficient quality for single‐crystal X‐ray diffraction analyses were obtained (Figure 2). Tetrabromide 5 crystallized in the orthorhombic space group Pnna with Z=4. The crystal packing is driven by halogen bonding between two bromides with d _{C‐Br⋅⋅⋅Br}=3.63 Å and dispersion interactions of the bromides with the aliphatic hydrogen (d _{C‐H⋅⋅⋅Br}=3.08–3.36 Å, Figure 2 a), forming two‐dimensional sheets.20 The distance between adjacent sheets is dominated by Br–π interactions (d _Br‐π=3.63 Å),21 and these layers are twisted by 16.0° with respect to each other (Figure 2 b, c). Tetraaryl hexahydropyrene 6 crystallized in the triclinic space group P‐1 with Z=2 (Figure 2 d). The molecules interact only by weak dispersion interactions of the peripheral tert‐butyl groups with the central naphthyl subunits (Figure 2 e).

Figure 2.

Single‐crystal X‐ray structures of tetrabromohexahydropyrene 5 (top) and tetraarylhexahydropyrene 6 (bottom) as stick models.

For the final cyclization by palladium‐catalyzed direct arylation under C−H activation, typical reaction conditions (PdCl₂(PCy₃)₂, DMAc, 200 °C, 48 h)22 were used to give the targeted TIP 2 in 29 % yield, along with the threefold‐cyclized product 8 in 20 % yield (Scheme 2). We also performed the cyclization using a wide range of conditions (e.g., various concentrations, solvents, temperatures, duration, alternative Pd sources, alternative bases), but did not obtain greater than 29 % yields for 2 nor a higher ratio between the yields for 2 and 8.

Scheme 2.

Synthesis of tetraindenopyrene 2 by C−H activation.

The two compounds can be distinguished by ¹H NMR spectroscopy. Whereas TIP 2 shows four clearly defined signals between δ=7.0 and 7.5 ppm (corresponding exactly to the previous report13), trindenopyrene 8 shows a more complex signal pattern consisting of 14 signals (two signals overlap at 7.8 and 7.7 ppm) between δ=6.9 and 8.2 ppm (Figure 3 bottom). Mass spectrometry shows a molecular ion peak for 2 of m/z 722.495 (m/z calcd for C₅₆H₅₀ ⁺: 722.391), which is two mass units smaller than that of triindenopyrene 8 (m/z calcd for C₅₆H₅₂ ⁺: 724.407 found: 724.484), consistent with the missing C−C bond.

Figure 3.

¹H NMR spectra (400 MHz) of a) TIP 2 and b) trindenopyrene 8 in CD₂Cl₂ at room temperature.

TIP 2 showed a strong concentration dependence (c=0.10–3.08 mm) of the chemicals shifts in the ¹H NMR spectra in CD₂Cl₂ (Figure 4), which is indicative of strong π–π‐stacking.23 At room temperature, protons H^b (Δδ=0.37 ppm) and H^c (Δδ=0.19 ppm) are more weakly influenced than H^a (Δδ=0.92 ppm) and H^d (Δδ=0.63 ppm), because H^b and H^c are sterically shielded by the adjacent tert‐butyl group against stacking.

Figure 4.

¹H NMR spectra (400 MHz) of TIP 2 in CD₂Cl₂ at concentrations between 0.10 and 3.08 mm at 293 K.

Assuming infinite π‐stacks, the averaged association K _E was determined to be 2.45×10³±0.77×10³  m ⁻¹ (ΔG=−19.0 kJ mol⁻¹) at 293 K by a least‐squares curve fitting of the infinite (isodesmic) association model (for details see the Supporting Information).23a, 24 This association constant is much higher than values reported, for example, for hexabenzocoronene‐based thiophene dendrimers (K _E up to 710 m ⁻¹ in CDCl₃)25 and within an order of magnitude of values reported for various perylene‐ and naphthalene bisimides.24d From measurements performed at temperatures ranging from 243 to 293 K, we determined ΔH=−15.8 kJ mol⁻¹ and ΔS=11.9 J mol⁻¹ K⁻¹ from a van't‐Hoff plot; this suggested that the aggregation in solution is driven by both enthalpy and entropy.26 The strong aggregation tendency is also reflected by a moderate solubility of 2 in CH₂Cl₂ of (4.00±0.63) mg mL⁻¹. Attempts to grow single crystals from various solvents produced needle‐shaped crystals, which unfortunately could not be structurally refined by X‐ray diffraction.

TIP 2 was revisited by theoretical calculations. Scott mentioned that the TIP has a bowl‐shaped structure.12, 13 However, only the bowl‐to‐bowl inversion barrier calculated by DFT (B3LYP/6‐31G*) was discussed, and no further details, such as geometrical parameters or energy levels of frontier molecular orbitals, were provided (see also the Discussion below).14 To obtain further insights into the structural details of TIP 2, dispersion‐corrected (D3)27 DFT methods (B3LYP/6‐311G(d,p)) were used to calculate molecular properties. It was found that the input geometry (MM2 optimized models) is crucial to the outcome of the DFT optimization. Starting from a planar input, the DFT optimization using ultra‐tight convergence criteria also produced a planar geometry. A frequency analysis of the result shows an imaginary frequency with 15.5i cm⁻¹, which is indicative of a transition state. A second optimization, this time performed with an already contorted input, resulted in a bowl‐shaped structure that no longer shows an imaginary frequency; this indicates that it is a realistic energy minimum (Figure 5). The bowl depth is 0.69 Å (if measured to the original 2,7‐positions of the pyrene) or 1.44 Å (maximum bowl depth) and thus similar to the bowl depth of dibenzorubicene (1.68 Å).10b The dispersion‐corrected calculations gave a greater bowl depth than those performed without the D3 correction term (1.18 Å); this result deviates by about 20 %(!), thus indicating that dispersion has a significant effect in stabilizing a contorted structure. The tert‐butyl groups do not contribute substantially to the curving by dispersion interactions (see the Supporting Information).

Figure 5.

Geometry‐optimized model (B3LYP/6–311G(d,p)) of the bowl‐shaped TIP 2. a) Top view. b) side view. AICD plots (HF/6–31+G(d)) of 2 from c) the top/concave side with the magnetic field pointing out of the paper plane and red arrows indicating the direction of the ring current. d) Side view. Canonical MOs of 2 (B3LYP/6–311G(d,p)) with the corresponding orbital energy. e) LUMO; f) HOMO. All calculations were corrected for dispersion (D3).

Based on the dispersion‐corrected model, the bowl‐to‐bowl inversion barrier was calculated to be 6.47 kJ mol⁻¹ (ΔG=11.9 kJ mol⁻¹, Figure 6). This is substantially higher than without dispersion correction (2.99 kJ mol⁻¹) and even higher than the previously published value (1.38 kJ mol⁻¹).13 Although the estimated inversion barrier is higher, it still means that TIP 2 fluctuates 51 billion times per second between the bowl‐shaped minima, much too fast to be determined by variable‐temperature NMR measurements. In comparison, corannulene shows an experimentally determined inversion rate of 200 000 per second at room temperature with a corresponding inversion barrier of Δ^≠ G=43±1 kJ mol⁻¹ (10.2±0.2 kcal mol⁻¹).2

Figure 6.

Inversion process of 2 via a planar transition state with the calculated difference in energy (black) and free enthalpy (red) derived from DFT calculations (B3LYP/6–311G(d,p)).

Additional information on the electronic structure of TIP 2 were obtained by AICD28 and NICS calculations (HF/6–31+G(d), Figure 5). The outer benzene rings (A) show typical aromatic character with comparable NICS(−1) and NICS(+1) values of −8.8 and −8.7. NICS(−1) is the concave and NICS(+1) the convex side of the bowl. Aromaticity is also observed for the D‐rings (NICS(−1)=−11.7 and NICS(+1)=−7.3, due to the higher electron density on the bowl's concave side. As known from unsubstituted pyrene,12 the C‐ring shows smaller aromaticity, with NICS(−1)=−5.3, and here the curvature has by far the greatest influence on the virtual chemical shift, with NICS(+1)=−0.6. The five‐membered B‐rings are nearly nonaromatic, with NICS(−1)=1.3 and NICS(+1)=3.7, and with a tendency toward antiaromatic character, similar to PAHs with fused five‐membered rings such as corannulene29 or others.30 The ring currents derived by AICD calculations (see red arrows in Figure 5 c) are in accordance with the trends observed by the NICS calculations.

The DFT‐calculated energies of the FMOs are E _{HOMO, DFT}=−5.5 eV and E _{LUMO, DFT}=−3.0 eV (Figure 5 e and f). Although no oxidation was recorded within the redox window of the solvents employed (CH₂Cl₂ and o‐DCB) at anodic potentials, two quasireversible reduction potentials were found in both these solvents (Figure 7). In CH₂Cl₂, the reduction potentials are ${\mathrm{E}}_{\mathrm{red},1}^{1/2}$ =−1.41 V and ${\mathrm{E}}_{\mathrm{red},2}^{1/2}$ =−1.74 V. In o‐DCB the two reduction peaks were found at slightly lower potentials (${\mathrm{E}}_{\mathrm{red},1}^{1/2}$ =−1.48 V and ${\mathrm{E}}_{\mathrm{red},2}^{1/2}$ =−1.84 V). The first reduction potentials are higher by about 0.3–0.4 V compared to [60]PCBM (${\mathrm{E}}_{\mathrm{red},1}^{1/2}$ =−1.08 V) and [70]PCBM (${\mathrm{E}}_{\mathrm{red},1}^{1/2}$ =−1.09 V).31 Making a commonly used assumption, the electron affinity can be estimated as EA=‐${\mathrm{E}}_{\mathrm{red},1}^{1/2}$ +4.8 eV),32 corresponding to EA=−3.46 eV in CH₂Cl₂ and EA=−3.42 eV in o‐DCB.

Figure 7.

Cyclic voltammograms of TIP 9 in CH₂Cl₂ (black) and o‐DCB (red), measured at room temperature with a Pt electrode, nBu₄NPF₆ (0.1 m) as electrolyte, and Fc/Fc⁺ as internal reference (scanning speed: 100 mVs⁻¹).

The calculated and experimentally determined FMOs suggest that TIP 2 is potentially interesting for organic electronics applications, both as an electron‐ and as a hole‐conducting semiconductor. Initial experiments using 2 in thin‐film transistors (TFTs) indicate hole mobilities of 4×10⁻⁴ cm² V⁻¹ s⁻¹ in TFTs fabricated on silicon substrates and 1×10⁻⁴ cm² V⁻¹ s⁻¹ in TFTs on flexible polyethylene naphthalate (PEN) substrates and on/off current ratios up to 10³ measured under ambient conditions (for details, see the Supporting Information).

In summary, we have introduced an alternative synthetic approach to achieve TIP 2 in five consecutive steps and with a 50‐fold higher yield both for the final cyclization step (29 vs. 0.5 %13) and for the overall synthesis (13 vs. 0.25 %). TIP 2 was revisited by dispersion‐corrected DFT calculations, revealing that the bowl‐to‐bowl inversion barrier is substantially higher than previously estimated. Furthermore, TIP 2 was used to fabricate p‐channel TFTs, indicating charge‐carrier mobilities up to 4×10⁻⁴ cm² V⁻¹ s⁻¹ and on/off current ratios of up to 10³. To the best of our knowledge, this is the first example of a transistor based on a nonfunctionalized hydrocarbon buckybowl.33, 34, 35 The possibility of easily scaling up the synthesis of precursor 3 without the necessity for purification by column chromatography will allow us to provide TIP 2 in sufficiently high amounts to explore its chemistry and physics; this is ongoing in our laboratory.

Experimental Section

For experimental details, see the Supporting Information.

Conflict of interest

The authors declare no conflict of interest.

Supporting information

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

S. M. Elbert, A. Haidisch, T. Kirschbaum, F. Rominger, U. Zschieschang, H. Klauk, M. Mastalerz, Chem. Eur. J. 2020, 26, 10585.