Anthraquinodimethane Ring-Flip in Sterically Congested Alkenes: Isolation of Isomer and Elucidation of Intermediate through Experimental and Theoretical Approach

Abstract

From the viewpoint of the development of molecular response systems, stimulus-induced switching of multi-conformational/multi-configurational overcrowded ethylenes are interesting, whose properties could be manipulated by understanding the detailed isomerization paths. Anthraquinodimethane (AQD) ring-flip is usually a very fast process, and thus less studied experimentally. Herein, we studied AQDs with dibenzo- and tribenzocycloheptatrienylidene units, which have large enough steric hindrance to retard the AQD ring-flip to allow determination of the ΔG^≠ value experimentally. Their thermal isomerization was also scrutinized using the artificial force induced reaction method to elucidate the intermediates. Based on the structural unsymmetry in a newly prepared AQD, one of the isomers that undergoes a reversible conformational change via AQD ring-flip was isolated and analyzed by X-ray for the first time.

1. Introduction

Bis(tricyclic) aromatic enes are representative overcrowded ethylenes (OCEs)¹ that exhibit changes in conformation in response to external stimuli, such as light, heat, and mechanical force, and thus serve as useful materials for electronic devices or molecular switches.^2–7 Only a few papers on OCEs have reported the successful isolation and X-ray analyses of multiple conformers/configurational isomers (e.g., I⁸ and II,⁹ Chart 1). When a quinoid-type skeleton is incorporated within the framework of an OCE, facile and reversible electron transfer can be induced to allow interconversion with the corresponding ionic states (e.g., II and III^10,11). Thus, well-designed quinoid-type OCEs are promising candidates for the development of molecular response systems for which redox behavior can be controlled by external stimuli.¹²

In this regard, 9,10-anthraquinodimethane (AQD) derivative (1) with two dibenzocycloheptatrienylidene units is noteworthy.¹³ Due to steric congestion, each of the exomethylene bonds is deformed into a folded form to provide anti- and syn-configurational isomers, and both the most stable anti,anti-1 (AA-1) (E_rel = 0 kcal mol⁻¹; ωB97X-D/6-31G*) and the second-most stable syn,anti-1 (SA-1) (E_rel = +2.42 kcal mol⁻¹) were isolated as stable entities, which do not interconvert under ambient conditions due to the semi-rigid molecular framework. There is no indication of formation of the syn,syn-1 (SS-1) (E_rel = +11.3 kcal mol⁻¹), which suffers from the largest steric repulsion among three. Upon electrochemical oxidation, both AA-1 and SA-1 were oxidized to the same dication (1²⁺), which was isolated as a persistent salt thanks to the aromaticity of dibenzotropylium units (Scheme 1). Notably, AA-1 and SA-1, have different oxidation potentials. Since two configurational isomers are quantitatively interconvertible upon UV-irradiation (one-way, from AA-1 to SA-1) and heating (one-way, from SA-1 to AA-1), switching of oxidative properties by external stimuli (light and heat) could be accomplished for the first time using OCE-type compounds. Similar behavior was observed for less-donating AQD 2¹³ with two tribenzocycloheptatrienylidene units, for which anti,anti-2 (AA-2) and syn,anti-2 (SA-2) were also interconvertible quantitatively by light and heat for selective oxidation.

Apparently, configurational change in only one of the two cycloheptatrienylidene units is responsible for the redox switching in 1 and 2. However, we independently found that mono(cycloheptatrienylidene) derivatives 4a, b do not afford the redox switching since they only exist as the anti-form (see Supporting Information), whereas the corresponding syn-configurational isomers (S-4a, b) could not be observed even under the photoirradiation of A-4a, b. It is most probable that as-generated S-4a, b (E_rel = +2.35 and +2.36 kcal mol⁻¹ for 4a and b, respectively; ωB97X-D/6-31G*) undergoes facile isomerization via “AQD ring-flip” to A-4a, b (E_rel = 0 kcal mol⁻¹) (Scheme 2), whose energy barrier is generally too low to stop in the butterfly-shaped 11,11,12,12-tetrasubstituted AQD derivatives.^14,15

We envisaged that the molecular framework in 1 and 2 is quite suitable to examine the AQD ring-flip due to the large steric hindrance around the double bonds, thus allowing us to scrutinize the reaction paths not only by a theoretical but also an experimental approach. The ring-flip would induce conformational isomerization of AA-1, 2 into SS-1, 2, whereas SA-1, 2 would undergo degenerate isomerization while exchanging the anti- and syn-parts.

This paper describes experimental and theoretical studies on the thermal ring-flip behavior of 1 and 2. The SS-isomers are energetically too high to be generated from AA-1, 2. On the other hand, the degenerate isomerization has been evidenced for SA-1, 2, and the intermediates and transition-states were elucidated by the aid of artificial force induced reaction (AFIR) method,^16–18 which has recently attracted attention as a promising computational method for finding reaction paths without prejudgment of the reaction coordinates. To further investigate the AQD ring-flip process experimentally, an unsymmetrical compound 3 was newly synthesized, which possesses both dibenzo- and tribenzocycloheptatrienylidene units. The non-degenerate ring-flip causes isomerization between SA-3 and AS-3, which would still be fast at ambient conditions, as in SA-1, 2. It should be noted that one of the isomers (AS-3) was successfully isolated and analyzed by X-ray, which represents the very first example among AQDs that undergo isomerization via the AQD ring-flip.

2. Results and Discussion

2.1 AQD Ring-Flip Process in Symmetric AQDs 1 and 2 with the Same Cycloheptatrienylidene Units

For symmetric AQDs 1 and 2 (R = R′),¹³ their thermal isomerization behaviors had been previously studied only in terms of the transformation of SA-1, 2 to the most stable isomer, AA-1, 2, which have activation energies of 30.5 and 40.9 kcal mol⁻¹, respectively. The isomerization process is independent from the AQD ring-flip, which would have much lower energy barriers. For these configurational isomers, the AQD ring-flip would cause the interconversion between AA-1, 2 and SS-1, 2, or degenerate interconversion of SA-1, 2 with exchange of syn- and anti-parts.

With a large energy difference between AA- and SS-isomers (11.3 kcal mol⁻¹ for 1 and 22.0 kcal mol⁻¹ for 2, ωB97X-D/6-31G*), the contributing amount of SS-isomer is limited, which agrees with the variable temperature (VT)-NMR analyses of AA-1, 2 at the elevated temperatures (Figures S15 and S16). Upon elevating the temperature, slight shift of resonances was observed, which might be related to the increased contribution from SS-1, yet the change is so subtle. Thus, we focus on the degenerate isomerization of SA-1, 2.

A VT-NMR analysis of bis(dibenzo) compound SA-1 in DMSO-d₆ showed the coalescence of signals as depicted in Figure 1. Thus, the anti- and syn-parts in SA-1 do exchange via the AQD ring-flip. Based on T_c (393 K) for the vinylic protons (H1) of the dibenzocycloheptatriene unit, the ΔG^≠ value of 19.4 kcal mol⁻¹ was determined, which is much smaller than that for the configurational isomerization from SA-1 to AA-1. Very similar behavior of AQD ring-flip was observed for bis(tribenzo) analogue SA-2 (ΔG^≠ = 19.2 kcal mol⁻¹, T_c = 393 K) (Figure S17). Close inspection of these spectra showed that there is an intriguing temperature-dependent shift of the resonances, suggesting that a certain intermediate, which has a set of chemical shifts different from that of the SA-form, would be present during the AQD ring-flip of SA-1, 2.

2.2 Intermediate for AQD Ring-Flip of SA-1 and SA-2

Paths for the experimentally observed AQD ring-flip (SA-AS transformation) were investigated theoretically for the case of bis(dibenzo) compound SA-1 using the AFIR method.^19,20 Figure 2 shows the energy variation along the intrinsic reaction coordinate (IRC) paths and molecular geometries on the critical points. Scheme 3 illustrates the molecular motion along the reaction path. Through the reaction, the syn- and anti-parts exchange with moderate barriers. This reversible process explains the experimentally observed degenerate thermal process. As shown in Scheme 3a, the geometrical change occurs sequentially first in the syn-part and then in the anti-part. There is a metastable intermediate called chair, which was newly found in this study. The reaction movie for the IRC path is available in the Supporting Information (Movie S1).

In the newly found conformer, chair-1, both of the dibenzocycloheptatrienylidene units are extended, and the central quinodimethane part adopts a unique chair form with both benzene rings on the quinodimethane unit located in the same plane (Figure 3a), which causes higher energy by 18.1 kcal mol⁻¹ than SA-1. Since chair-1 has a centrosymmetric structure, two dibenzocycloheptatrienylidene units are identical, and thus this conformer could be the intermediate for exchanging the anti- and syn-parts of SA-1. The transition state toward chair-1 has a relative energy of 21.5 kcal mol⁻¹, which is similar to the observed energy barrier determined by VT-NMR.

Although we found another path involving the less stable intermediate chair-1′ with folded dibenzocycloheptatrienylidene units (Figure 2b), in which the geometrical change occurs sequentially first in the anti-part and then in the syn-part (Scheme 3b), the path is less likely as the AQD ring-flip process due to its higher barriers (26.9 kcal mol⁻¹). The metastable intermediate (chair-1′) also has a higher energy (25.6 kcal mol⁻¹) than chair-1.

Similar paths were studied also for bis(tribenzo) compound SA-2, and paths containing intermediates chair-2 and chair-2′, respectively, were obtained (ΔE = 17.5 kcal mol⁻¹ for chair-2 and 20.7 kcal mol⁻¹ for chair-2′, respectively) (Figure S21, Movies S2). Again, paths through folded intermediate chair-2′ were found to be energetically unfavorable. Thus, AQD ring-flip of SA-2 proceeds via chair-2 (Figure 3b) with first motion of the syn-part.

2.3 Molecular Design and Preparation of Unsymmetrically Substituted AQD 3

The above experimental and theoretical examinations have shown that the AQD ring-flip occurs easily even for the sterically congested compounds 1 and 2, which led us to design a compound with both dibenzo- and tribenzocycloheptatrienylidene units (3). As in the case of 1 and 2, we would get less information on AQD ring-flip between AA-3 (E_rel = 0 kcal mol⁻¹; ωB97X-D/6-31G*) and SS-3 (E_rel = +12.8 kcal mol⁻¹) due to very small contribution of SS-3. However, there would be two kinds of syn,anti-forms [SA-3 (E_rel = +2.63 kcal mol⁻¹) and AS-3 (E_rel = +1.06 kcal mol⁻¹)] due to the two different units on the exomethylene bonds, which would interconvert via the non-degenerate AQD ring-flip (Scheme 4). In this way, the unsymmetric system would provide a chance to study AQD ring-flip while isolating the pure sample of isomers (SA-3 and/or AS-3).

Previously, AQDs 1 and 2 were obtained as isomer mixtures of AA- and SA-forms upon reduction of dicationic salts 1²⁺(BF₄⁻)₂ and 2²⁺(BF₄⁻)₂, respectively, which in turn were prepared from 9,10-dibromoanthracene and dibenzo- or tribenzocycloheptatrienone (Scheme 5a).¹³ In this work, to prepare the unsymmetrically substituted compound selectively without contamination by symmetric counterparts, a stepwise strategy was adopted (Scheme 5b), which includes the Barton-Kellogg protocol^21,22 in the last stage. Another attractive point of this scheme is the selective formation of a syn-type folded alkene unit⁵ during desulfurization of the episulfide precursor, which is favorable for studying AQD ring-flip between SA-3 and AS-3.

Thus, 9-lithio-10-methoxyanthracene was generated from 9-bromo-10-methoxyanthracene²³ and reacted with tribenzocycloheptatrienone 5²⁴ to give tert-alcohol 6 in 55% yield. By subsequent treatment with acid and then base, 6 was converted to 10-tribenzocycloheptatrienylidene-9-anthrone 7 in 52% yield, which was in turn converted to thione 8 by Lawesson's reagent in 92% yield. Reaction of 8 with 5-diazodibenzocycloheptatriene 9⁵ gave intermediate episulfide 10, which underwent in situ desulfurization to give desired unsymmetrically substituted AQD 3 quantitatively.

Upon recrystallization of as-prepared 3, single-crystalline specimens were obtained, which were suitable for X-ray analysis (Figure 4). The ORTEP drawings clearly show that the molecule in the crystal is AS-3, in which the dibenzocycloheptatrienylidene part adopts an anti-form while the tribenzocycloheptatrienylidene unit has a syn-form with the most outer benzene ring bending deeply into the central AQD unit. In this way, we successfully isolated one of the isomers, even though it would undergo facile isomerization via the AQD ring-flip in solution.

2.4 AQD Ring-Flip Process in Unsymmetrically Substituted AQD SA-3 and AS-3

The ¹H NMR spectrum obtained by dissolving pure crystals of AS-3 indicated coexistence of a minor amount of SA-3, which would be generated from AS-3 by the facile AQD ring-flip in solution. No AA-3 was present, as expected. SA-3 and AS-3 are in a 1:3 ratio when measured in CDCl₃ or CD₂Cl₂ at 298 K. The isomer ratio was proven to be solvent-dependent (SA-3:AS-3 = 1:4 in C₆D₆ and 1:5 in DMSO-d₆, respectively), and the equilibrium is attained instantaneously. The preference of AS-3 in equilibrated mixture can be accounted for by the calculated energy difference of 1.57 kcal mol⁻¹ in favor of AS-3 (Table S2). We assumed that tribenzocycloheptatrienylidene skeleton intrinsically has more severe steric strain than dibenzo skeleton even in the anti-form, and thus the unsymmetrical compound preferred AS-3 to avoid the energy loss from syn-folded dibenzocycloheptatrienylidene unit.

Further evidence for AQD ring-flip between SA-3 and AS-3 was obtained by a VT-NMR analysis showing the coalescence of signals of SA-3 and AS-3 upon heating in DMSO-d₆ to give a ΔG^≠ value of 19.3 kcal mol⁻¹ (T_c = 378 K) (Figure 5). Using the AFIR method, the intermediate (chair-3) is elucidated, which is higher in energy by 18.4 kcal mol⁻¹ than AS-3. The energy barriers to reach the intermediate are 20.7 and 21.0 kcal mol⁻¹, respectively, from SA-3 and AS-3 (Figure S22) with motion of the syn-part first (Movie S3). Another calculated intermediate chair-3′, which would be generated by first motion of the anti-part, is less stable as in the case of 1 and 2, and thus would not be involved in AQD ring-flip between SA-3 and AS-3. Although the calculations were conducted in unrestricted condition by considering the possibility of open-shell species during AQD ring-flip, we could not find any contributions of radical species in any derivatives.

2.5 Photochemical and Other Thermal Processes in Unsymmetrically Substituted AQD 3

Through the synthetic route shown in Scheme 5b, the most stable isomer AA-3 was not obtained, whereas it was obtained in 98% yield when a mixture of SA-3 and AS-3 was heated at reflux for 4 h in DMSO. When the reaction was followed by NMR, clean conversion to AA-3 was observed (Figure 6). The activation energy (E_a = 30.8 kcal mol⁻¹) for the thermal isomerization of SA/AS-3 to AA-3 was determined by Arrhenius plot based on the rate constants and the thermodynamic parameters (ΔH^≠ = 30.0 kcal mol⁻¹ and ΔS^≠ = −9.58 cal mol⁻¹) were estimated by the Eyring equation (Table S1 and Figure S18). The value of E_a is close to that of 1 (30.5 kcal mol⁻¹) and rather different from that of 2 (40.9 kcal mol⁻¹), suggesting that the transition state of the reaction from SA/AS-3 to AA-3 is similar to that of SA-1 to AA-1, which would give valuable information upon clarifying the reaction mechanism in the future.

On the other hand, selective photoirradiation of AA-3 in CH₂Cl₂ with a 370-nm light caused clean isomerization of one of the cycloheptatriene units from an anti-form to a syn-form. Due to rapid AQD ring-flip, the product is a mixture of SA-3 and AS-3 (Figure 7). In a preparative-scale experiment, photoreaction proceeded smoothly to give the isomer quantitatively (500 W Xe, λ > 360 nm, 1.26 mM in CH₂Cl₂). In this way, the newly synthesized unsymmetrically substituted AQD 3 inherits the essential thermal/photochemical reactivities from 1 and 2 while providing some degeneration of isomerism.

3. Conclusion

Almost all of the 11,11,12,12-tetrasubstituted AQDs adopt a butterfly-shaped geometry to relieve the steric hindrance between the peri-hydrogens and the substituents on the exocyclic bonds. The central hexagon is deformed into a deeply folded boat-form whereas its ring-flip is usually very fast. For the AQD derivatives 1–3 with both exocyclic bonds being OCE due to rigid and bulky dibenzo-/tribenzocycloheptatrienylidene units, the AQD ring-flip is retarded, thus providing a suitable platform to study the AQD ring-flip in detail. Experimentally, the ΔG^≠ values around 20 kcal mol⁻¹ were determined for the degenerate isomerization of SA-1, 2, in which one exomethylene is bent to a syn-form and the other one into an anti-form with mutual exchange of syn- and anti-units via AQD ring-flip. For the newly synthesized unsymmetrically substituted AQD 3, exchange of syn- and anti-units via AQD ring-flip causes non-degenerate isomerism between two configurational isomers (SA-3 and AS-3), for which AS-3 was successfully isolated and its crystal structure was analyzed by X-ray. Reaction path calculations using the AFIR method revealed that the AQD ring-flip starts with the motion of the syn-cycloheptatrienylidene unit to generate the intermediate (chair-1–3), in which the central hexagon of AQD adopts a chair-form. Another possibility is the first motion of the anti-cycloheptatrienylidene unit to generate the chair-type central hexagon, however, this process may lead the different intermediate (chair-1′–3′) with a higher energy, thus less probable as the reaction path of the AQD ring-flip. The detailed reaction mechanism revealed in this study would help to understand the phenomena involving several conformers/configurational isomers in the semi-rigid framework.²⁵ OCEs exhibiting stimulus-induced switching behavior should attract further attention for the development of electronic devices or molecular switches, and a similar approach based on both experimental and theoretical considerations can serve as an essential tool for better understanding semi-rigid functional molecules.

4. Experimental Section

General

All reactions were carried out under an argon atmosphere. Column chromatography was performed on silica gel 60 N (KANTO KAGAKU, spherical neutral) of particle size 40–50 µm or Wakogel^® 60 N (neutral) of particle size 38–100 µm. ¹H and ¹³C NMR spectra were recorded on a BRUKER Ascend^TM 400 (¹H/400 MHz and ¹³C/100 MHz) spectrometer. IR spectra were measured on a Shimadzu IRAffinity-1S spectrophotometer using the attenuated total reflection (ATR) mode. Mass spectra were recorded on a JMS-T100GCV spectrometer in FD mode by Dr. Eri Fukushi and Mr. Yusuke Takata (GC-MS & NMR Laboratory, Research Faculty of Agriculture, Hokkaido University). Melting points were measured on a Yamato MP-21 and are uncorrected. UV-vis spectra were recorded on a Hitachi U-2910 spectrophotometer. Fluorescence spectra were measured on a Hitachi F-7000 spectrofluorometer. Fluorescence quantum yields were determined by using 9,10-diphenylanthracene (Φ_F = 0.97) as an external standard.²⁶ For photoisomerization reaction, a 150 W Xe lamp in the Hitachi F-7000 spectrofluorometer was used for an NMR-tube and an Ushiospax SX-UID501XAMQ light source device (500 W Xe lamp) was used with a CORNING COLOR FILTER (No. O-51) in a preparative scale. DFT calculations were performed with the Gaussian 16 program package.²⁷

9-(5H-Dibenzo[a,d]cycloheptatrien-5-ylidene)-10-(9H-tribenzo[a,c,e]cycloheptatrien-9-ylidene)-9,10-dihydroanthracene (syn,anti-isomers) (SA-3 and AS-3)

A solution of thione 8 (545 mg, 1.21 mmol) and diazo compound 9 (396 mg) in dry THF (15 mL) was heated to reflux for 13 h. After cooling to 25 °C, the resulting reaction mixture was concentrated under reduced pressure. The crude product was purified by column chromatography on silica gel (CH₂Cl₂/hexane = 1/1) to give a mixture of SA-3 and AS-3 (734 mg) as a white solid in 100% yield.

Mp: >300 °C; ¹H NMR (CDCl₃): δ/ppm SA-3: 7.85 (2H, d, J = 8.0 Hz), 7.76 (2H, dd, J = 3.4 Hz, 5.8 Hz), 7.57 (2H, d, J = 7.7 Hz), 7.51–7.48 (2H, m), 7.32–7.20 (12H, m), 7.03–7.00 (2H, m), 6.98 (2H, s), 6.76 (2H, d, J = 7.7 Hz), 6.62 (2H, t, J = 7.6 Hz), 6.42 (2H, d, J = 7.6 Hz); AS-3: 7.89 (2H, d, J = 7.6 Hz), 7.68 (2H, d, J = 8.5 Hz), 7.66 (2H, dd, J = 3.4 Hz, 5.8 Hz), 7.50 (2H, t, J = 6.6 Hz), 7.43 (2H, t, J = 7.6 Hz), 7.32–7.20 (6H, m), 7.16 (2H, t, J = 7.4 Hz), 7.03–7.00 (4H, m), 6.81 (2H, t, J = 8.5 Hz), 6.57 (2H, t, J = 8.5 Hz), 6.22 (2H, d, J = 7.4 Hz), 6.19 (2H, d, J = 8.5 Hz); ¹³C NMR (CDCl₃): δ/ppm 143.90, 143.08, 139.25, 138.84, 138.82, 138.63, 138.50, 137.42, 137.25, 136.98, 136.66, 136.53, 136.49, 136.13, 135.80, 135.13, 134.79, 134.75, 133.90, 133.86, 133.48, 131.27, 130.82, 130.13, 129.82, 129.45, 129.37, 129.02, 128.95, 128.91, 128.24, 127.95, 127.93, 127.86, 127.82, 127.71, 127.65, 127.62, 127.44, 127.37, 127.36, 127.03, 126.91, 126.78, 126.49, 126.41, 126.37, 125.31, 125.15, 125.07, 125.04, 124.99; IR (ATR): ν/cm⁻¹ 3058, 3018, 1484, 1473, 1449, 1430, 1214, 1160, 1111, 1098, 1041, 946, 894, 881, 858, 834, 801, 792, 768, 743, 720, 707, 668, 645, 624, 616, 606, 522, 462; LR-MS (FD) m/z (%): 609.30 (5), 608.29 (16), 607.29 (54), 606.29 (M⁺, bp); HR-MS (FD) Calcd. for C₄₈H₃₀: 606.23475; Found: 606.23508; UV-Vis (CH₂Cl₂): λ_max/nm (ε/L mol⁻¹ cm⁻¹) 232 (75700); Fluorescence (CH₂Cl₂, λ_ex = 290 nm): λ_em/nm (Φ_F) 410 (0.30).

Thermal Isomerization of SA-3 and AS-3 to AA-3

A solution of a mixture of SA-3 and AS-3 (180 mg, 296 µmol) in dimethylsulfoxide (10 mL) was heated to reflux for 4 h, and then diluted with water after cooling to 23 °C. The whole mixture was extracted with CH₂Cl₂ three times. The combined organic layers were washed with water and brine, and dried over anhydrous Na₂SO₄. After filtration, the solvent was concentrated under reduced pressure. The crude product was purified by column chromatography on silica gel (CH₂Cl₂/hexane = 1/1) to give AA-3 (176 mg) as a white solid in 98% yield.

Mp: >300 °C; ¹H NMR (CDCl₃): δ/ppm 7.84 (2H, dd, J = 3.4 Hz, 5.8 Hz), 7.67 (2H, dd, J = 1.0 Hz, 7.8 Hz), 7.54 (2H, dd, J = 3.4 Hz, 5.8 Hz), 7.45 (2H, dd, J = 1.0 Hz, 7.5 Hz), 7.35 (2H, dt, J = 1.3 Hz, 7.8 Hz), 7.31–7.21 (6H, m), 7.19 (2H, s), 7.14 (2H, dd, J = 1.0 Hz, 4.9 Hz), 7.12 (2H, dd, J = 1.4 Hz, 5.2 Hz), 6.67–6.57 (6H, m), 6.41 (2H, dd, J = 1.6 Hz, 7.7 Hz); ¹³C NMR (CDCl₃): δ/ppm 144.12, 139.31, 139.25, 138.66, 137.54, 137.18, 137.11, 137.03, 135.64, 134.79, 133.59, 131.50, 129.85, 129.21, 129.13, 128.60, 128.20, 128.00, 127.96, 127.87, 127.76, 127.59, 127.31, 126.84, 125.08, 125.04; IR (ATR): ν/cm⁻¹ 3064, 3014, 1927, 1594, 1481, 1461, 1450, 1432, 1292, 1270, 1152, 1109, 1040, 1002, 948, 895, 836, 799, 792, 787, 773, 760, 750, 741, 721, 705, 644, 622, 617, 602, 590, 470; LR-MS (FD) m/z (%): 609.28 (5), 608.27 (17), 607.27 (54), 606.27 (M⁺, bp); HR-MS (FD) Calcd. for C₄₈H₃₀: 606.23475; Found: 606.23576; UV-Vis (CH₂Cl₂): λ_max/nm (ε/L mol⁻¹ cm⁻¹) 292 (26200), 232 (85500); Fluorescence (CH₂Cl₂, λ_ex = 290 nm): λ_em/nm (Φ_F) 406 (0.24).

Photochemical Isomerization of AA-3 to SA-3 and AS-3

A solution of AA-3 (45.7 mg, 75.3 µmol) in CH₂Cl₂ (60 mL, 1.26 mmol L⁻¹) was degassed by Ar bubbling, and then stirred at 25 °C for 14 h under photoirradiation by a 500 W Xe lamp with a CORNING (No. O-51) cut filter (λ > 360 nm). The solvent was concentrated under reduced pressure to give a mixture of AS-3 and SA-3 (46.0 mg) as a white solid in 100% yield.

9-(10-Methoxyanthracen-9-yl)-9H-tribenzo[a,c,e]cycloheptatrien-5-ol (6)

To a solution of 9-bromo-10-methoxyanthracene (304 mg, 1.06 mmol) in dry THF (5 mL) was added ⁿBuLi (1.58 mol L⁻¹ in hexane, 0.80 mL, 1.26 mmol) dropwise over 2 min at −78 °C. After stirring at −78 °C for 1 h, tribenzocycloheptatrienone 5 (320 mg, 1.25 mmol) was added to the suspension, and the mixture was warmed to 24 °C. The resulting solution was stirred at 24 °C for 1 h, and then diluted with water. The whole mixture was extracted with CH₂Cl₂ three times. The combined organic layers were washed with water and brine, and dried over anhydrous Na₂SO₄. After filtration, the solvent was concentrated under reduced pressure. The crude product was purified by column chromatography on silica gel (CH₂Cl₂/hexane = 1/3) to give 6 as a light yellow solid (269 mg) in 55% yield.

Mp: 248–252 °C (decomp.); ¹H NMR (CDCl₃): δ/ppm, 7.96 (2H, dd, J = 1.4 Hz, 8.8 Hz), 7.71 (2H, dd, J = 1.0 Hz, 9.2 Hz), 7.52 (2H, dt, J = 1.3 Hz, 7.6 Hz), 7.43 (2H, dd, J = 1.3 Hz, 7.6 Hz), 7.21 (2H, dt, J = 1.3 Hz, 7.6 Hz), 7.16 (2H, ddd, J = 1.0 Hz, 6.4 Hz, 8.8 Hz), 7.05 (2H, dd, J = 1.3 Hz, 7.6 Hz), 6.85 (2H, ddd, J = 1.4 Hz, 6.4 Hz, 9.2 Hz), 6.65 (2H, dd, J = 2.4 Hz, 3.4 Hz), 6.39 (2H, dd, J = 2.4 Hz, 3.4 Hz), 3.85 (3H, s), 3.41 (1H, s); ¹³C NMR (CDCl₃): δ/ppm 152.24, 150.60, 138.64, 134.55, 130.74, 129.75, 129.59, 127.71, 126.30, 126.12, 125.82, 125.68, 124.05, 123.57, 123.48, 121.66, 121.14, 79.46, 62.04; IR (ATR): ν/cm⁻¹ 3568, 3548, 3480, 3058, 3016, 2938, 2860, 2833, 1809, 1619, 1556, 1519, 1484, 1472, 1437, 1403, 1359, 1315, 1274, 1168, 1157, 1125, 1088, 1056, 1026, 993, 969, 958, 849, 775, 766, 744, 738, 706, 679, 648, 617; LR-MS (FD) m/z (%): 466.20 (8), 465.20 (40), 464.20 (M⁺, bp); HR-MS (FD) Calcd. for C₃₄H₂₄O₂: 464.17763; Found: 464.17851.

10-(9H-Tribenzo[a,c,e]cycloheptatrien-9-ylidene)-9(10H)-anthrone (7)

To a solution of 6 (167 mg, 0.360 mmol) in CH₂Cl₂ (6 mL) were added trifluoroacetic acid (250 mg, 2.19 mmol) and H₂O (70 mg, 3.89 mmol) at 24 °C. After stirring at 24 °C for 17 h, the mixture was cooled to 0 °C. To the solution was added 20% NaOH aq. (5 mL) at 0 °C. The resulting mixture was stirred at 0 °C for 5 min, and then diluted with water. The whole mixture was extracted with CH₂Cl₂ three times. The combined organic layers were washed with water and brine, and dried over anhydrous Na₂SO₄. After filtration, the solvent was concentrated under reduced pressure. The crude product was filtered, washed with ethanol three times, and dried in vacuo to give 7 (81.1 mg) as a light yellow solid in 52% yield.

Mp: 266–268 °C; ¹H NMR (CDCl₃): δ/ppm 8.12 (2H, d, J = 7.6 Hz), 7.86 (2H, dd, J = 3.4 Hz, 5.7 Hz), 7.66 (2H, d, J = 7.6 Hz), 7.58 (2H, dd, J = 3.4 Hz, 5.7 Hz), 7.33 (2H, t, J = 7.6 Hz), 7.29 (2H, t, J = 7.6 Hz), 7.11 (2H, t, J = 7.8 Hz), 7.07 (2H, t, J = 7.8 Hz), 6.78 (4H, d, 7.8 Hz); ¹³C NMR (CDCl₃): δ/ppm 185.89, 143.74, 143.07, 138.87, 138.56, 137.95, 133.18, 130.33, 129.64, 129.41, 129.31, 128.70, 128.13, 127.99, 127.64, 127.40, 126.51, 126.39; IR (ATR): ν/cm⁻¹ 3060, 3017, 1669, 1616, 1594, 1490, 1478, 1465, 1456, 1449, 1430, 1302, 1283, 1267, 1225, 1213, 1169, 1158, 1149, 1092, 1053, 1040, 1001, 983, 959, 945, 930, 881, 867, 811, 782, 776, 770, 756, 749, 740, 707, 702, 689, 658, 634, 616; LR-MS (FD) m/z (%): 434.19 (7), 433.18 (37), 432.18 (M⁺, bp); HR-MS (FD) Calcd. for C₃₃H₂₀O: 432.15141; Found: 432.15033.

10-(9H-Tribenzo[a,c,e]cycloheptatrien-9-ylidene)-10H-anthracene-9-thione (8)

A solution of ketone 7 (653 mg, 1.51 mmol) and 2,4-bis(4-methoxyphenyl)-1,3-dithia-2,4-diphosphetan-2,4-disulfide (Lawesson’s reagent) (342 mg, 847 µmol) in dry toluene (15 mL) was heated to reflux for 4 h. After cooling to 25 °C, the resulting green solution was concentrated under reduced pressure. The crude product was purified by column chromatography on silica gel (CH₂Cl₂/hexane = 1/1) to give 8 (643 mg) as a green solid in 94% yield.

Mp: 250–252 °C; ¹H NMR (CDCl₃): δ/ppm 8.12 (2H, d, J = 7.3 Hz), 7.86 (2H, dd, J = 3.4 Hz, 5.8 Hz), 7.65 (2H, d, J = 7.6 Hz), 7.57 (2H, dd, J = 3.4 Hz, 5.8 Hz), 7.31 (2H, t, J = 7.6 Hz), 7.19 (2H, t, J = 7.3 Hz), 7.09 (2H, t, J = 7.6 Hz), 7.07 (2H, t, J = 7.3 Hz), 6.79 (2H, d, J = 7.6 Hz), 6.74 (2H, d, J = 7.3 Hz); ¹³C NMR (CDCl₃): δ/ppm 222.77, 143.79, 143.20, 141.20, 138.88, 137.81, 132.21, 130.36, 129.91, 129.69, 129.26, 128.66, 128.07, 127.96, 127.75, 127.65, 127.27, 126.86; IR (ATR): ν/cm⁻¹ 3057, 3012, 2922, 2362, 1671, 1592, 1567, 1476, 1464, 1430, 1312, 1296, 1277, 1266, 1235, 1161, 1098, 1059, 1035, 1002, 975, 951, 944, 903, 845, 780, 773, 760, 744, 731, 703, 694, 670, 646, 619, 612, 594; LR-MS (FD) m/z (%): 450.15 (11), 449.15 (40), 448.15 (M⁺, bp); HR-MS (FD) Calcd. for C₄₈H₃₀S: 448.12857; Found: 448.12942.

X-ray Analysis of AS-3

Data were collected with a Rigaku XtaLAB Synergy (Cu-Kα radiation, λ = 1.54184 Å). The structure was solved by ShelXT (Sheldrick) and refined by the full-matrix least-squares method on F² with anisotropic temperature factors for non-hydrogen atoms. All the hydrogen atoms were located at the calculated positions and refined with riding. Olex2²⁸ was used for all calculations. Crystals were obtained by recrystallization from CHCl₃/hexane. MF: C₄₈H₃₀, FW: 606.72, colorless plate, 0.30 × 0.10 × 0.05 mm³, monoclinic I2/m, a = 9.84061(10) Å, b = 15.57506(18) Å, c = 25.1815(4) Å, β = 93.8959(12)°, V = 3850.61(8) ų, ρ (Z = 4) = 1.047 g cm⁻³. A total 12453 reflections were measured at T = 150 K. Numerical absorption correction was applied (µ = 0.450 mm⁻¹). The final R1 and wR2 values are 0.0515 [I ≥ 2σ(I)] and 0.1749 (all data) for 4075 reflections and 223 parameters. Estimated standard deviations are 0.0018–0.004 Å for bond lengths and 0.09–0.18° for bond angles. Solvent mask procedure was used for the analysis. CCDC 2096394.

Computational Study

In the reaction path calculations, all the minima and first-order saddles were optimized using the UωB97X-D functional and the 6-31G* basis set. The solvent effects of DMSO were accounted for by the conductor-like polarizable continuum model (CPCM).^29,30 The computational level is denoted by UωB97X-D/6-31G*. Approximate structures explored using the AFIR method were reoptimized to actual minima and first-order saddles without artificial force. The obtained first-order saddles were verified to connect two desired minima by IRC calculations.

In this study, all the energy, gradient and Hessian calculations were performed using the Gaussian 16 program,²⁷ and the AFIR searches were carried out using a developer version of the GRRM program.³¹

Acknowledgment

The results in this work are the outcomes achieved through the MANABIYA (ACADEMIC) program conducted by Institute for Chemical Reaction Design and Discovery (ICReDD), Hokkaido University, which was established by World Premier International Research Initiative (WPI), MEXT, Japan. We also thank Grant-in-Aid from MEXT and JSPS (Nos. 20H02719, 20K21184, 21H01912, and 21H05468) Japan. Y.I. acknowledges Toyota Riken Scholar, Hattori Hokokai, and the 2020 DIC Award in Synthetic Organic Chemistry, Japan. T.T. and Y.Hay. are grateful for the Ministry of Education, Culture, Sports, Science and Technology through the Program for Leading Graduate Schools (Hokkaido University “Ambitious Leader’s Program”). This work was also supported by the Research Program of “Five-star Alliance” in “NJRC Mater. & Dev.” MEXT. A part of the results was computed at the computer center of Kyoto University.

Supporting Information

¹H and ¹³C NMR spectra, theoretical calculations, and additional data for structural change are shown in the Supporting Information. This material is available on https://doi.org/10.1246/bcsj.20210355.

References

Yusuke Ishigaki

Yusuke Ishigaki received his B.S. degree in 2008, M.S. degree in 2009, and Ph.D. degree in 2012 from Hokkaido University. After working as a JSPS postdoctoral fellow at Ulm University (Professor Peter Bäuerle) and at Nippon Steel & Sumikin Chemical Co., Ltd., he moved to Hokkaido University as an assistant professor in 2016 and was appointed as an associate professor in 2021. His research interests are flexible C–C covalent bonds such as long C–C single bonds (hyper covalent bond) and strained C=C double bonds based on redox-active molecules.

Satoshi Maeda

Satoshi Maeda received his B.Sc. (2002) and PhD (2007) from Tohoku University (Prof. Koichi Ohno). During 2007–2010, he served as JSPS research fellow under the guidance of Prof. Ohno and Prof. Keiji Morokuma. In 2010–2012, he served as Assistant Professor of the Hakubi project at Kyoto University in the group of Prof. Morokuma. In 2012–2017, he became Assistant Professor then Associate Professor at Hokkaido University in the group of Prof. Tetsuya Taketsugu. He now is a full professor at Hokkaido University and the director of WPI-ICReDD.

Takanori Suzuki

Takanori Suzuki graduated from Tohoku University in 1984 and received his Ph.D. in 1988 under the supervision of Professors Toshio Mukai and Tsutomu Miyashi. After he worked at Tohoku University as a JSPS post-doctoral fellow for 1988–1989 and as Research Associate for 1989–1994, he moved to Faculty of Science, Hokkaido University and joined Professor Takashi Tsuji's group as Associate Professor (1995–2002). He was appointed as Professor at Hokkaido University in 2002.