*2.3. Cycloadditions of Benzyne to 2-Phenylindene*

Intrigued by the unexpected cycloaddition of benzyne to the indenyl substituent of 9-(3-indenyl)anthracene to form the methano-dihydrophenanthrene, 10, we chose to investigate the reaction of benzyne with 2-phenylindene. Although the majority of the starting indene was recovered, two products were isolated and fully characterised as benzyne adducts [11]. The first was readily identified spectroscopically as the known indeno-phenanthrene, 20, resulting from the addition of a single benzyne to 2-phenylindene. Presumably, the initially formed dihydrophenanthrene, 21, was further oxidised in the presence of excess isoamyl nitrite (Scheme 9).

**Scheme 9.** Cycloaddition reactions of benzyne to 2-phenylindene.

The formula of the second product corresponded to the addition of two benzyne units to 2-phenylindene; one can envisage the first step as the [4 + 2] cycloaddition to the indenyl moiety to form 22, entirely analogous to the reaction of benzyne with 9-(3-indenyl)anthracene to form 10. In the second step, [2 + 2] addition yields a cyclobutene, 23, that adopts the *syn*, rather than the *anti*, configuration because of the presence of the adjacent phenyl substituent. Subsequent thermolysis can open up the four-membered ring and bring about rearrangement to the observed product 24 that was unequivocally characterised by X-ray crystallography (Figure 10). The thermodynamic driving force for such a process would be the relief of steric strain in the cyclobutene ring, and the recovery of aromatic character in the original six-membered ring of the indene [11].

**Figure 10.** Structure of 13a-phenyl-13,13a-dihydro-8b*H*-indeno-[1,2-*l*]-phenanthrene, **24**.

#### *2.4. Organometallic Derivatives of Indenyl Anthracenes*

The palladium-mediated ready availability of 2-phenylindene, 9-(2-indenyl)anthracene and 9-(3-indenyl)anthracene, as well as their corresponding triptycenes, prompted an investigation of the syntheses, structures and dynamic behaviour of some of their organometallic derivatives.

The 2-phenylindene reacts with Cr(CO)6 in 1,4-dioxane at 125 ◦C to form two isomeric complexes wherein the Cr(CO)3 tripod is attached in an η<sup>6</sup> fashion either to the phenyl substituent, as in 25, or to the six-membered ring of the indene, as in 26 (Scheme 10). (η5-2-phenylindenyl)Re(CO)3, 27, was prepared either directly by heating the ligand with Re2(CO)10 in decalin in a sealed tube at 160 ◦C or indirectly by transmetallation of 2-phenyl-1-trimethylstannylindene. The structures of 25 and 27 appear as Figure 11 and show that the indenyl and phenyl planes differ only slightly from coplanarity [20].

**Scheme 10.** Chromium and rhenium tricarbonyl complexes of 2-phenylindene.

We note that complexes 25 and 27 can each adopt a mirror-symmetric (*C*s) conformation; in the chromium case, the indenyl unit must be aligned orthogonal to the plane of the phenyl ring (dihedral angle of 90◦), whereas, in the rhenium complex, 27, the molecular mirror plane only bisects both the indenyl and phenyl rings when they are coplanar (dihedral angle of 0◦). In all these cases, the rotation barrier about the axis connecting the rings is very low; DFT calculations suggest values in the range of 20–30 kJ mol−<sup>1</sup> [20].

**Figure 11.** Molecular structures of [η6-(2-indenyl)benzene]tricarbonylchromium, **25**, and (η5-2 phenylindenyl)tricarbonylrhenium, **27**.

As shown in Scheme 11, the syntheses of the chromium and rhenium tricarbonyl complexes of 9-(2-indenyl)anthracene, 28 and 29, respectively, parallel those of their 2-phenylindene counterparts. However, the considerably larger wingspan of the anthracenyl moiety brings about noticeable changes in their structures and dynamic behaviour; the interplanar indenyl anthracenyl angles in the chromium and rhenium complexes were found to be 62◦ and 52◦, respectively (Figure 12).

In the η6-bonded chromium system, 28, the rotational barrier for equilibration of the benzo rings of the anthracene was measured by V-T NMR as ~63 kJ mol−1, a value 15 kJ mol−<sup>1</sup> higher than that found for the (η5-indenyl)rhenium complex, 29, despite the Cr(CO)3 tripod being further away from the anthracene. However, one should recall that the NMR experiment yields a value for the energy separation between the ground state and the transition state for the dynamic process. In this case, rather than invoking a decrease in the energy of the transition state, it is more likely that the ground state

has been raised by steric interactions with the methylene hydrogen atoms rather than with the metal carbonyl tripodal fragment, thus lowering the observed barrier. It is, therefore, particularly noteworthy that upon deprotonation of [η6-2-(9-anthracenyl)indene]tricarbonylchromium, 28, which brings about a haptotropic shift of the tripod to form the anion [η5-2-(9-anthracenyl)indenyl]tricarbonylchromium, 30, closely analogous to the rhenium complex, 29, this rotational barrier has once again decreased, an effect we designated as an "anti-braking" phenomenon.

**Scheme 11.** Cr and Re carbonyl complexes of 2-indenyl- and 3-indenyl-anthracene.

**Figure 12.** Molecular structures of [η6-2-(9-anthracenyl)indene]tricarbonylchromium, **28**, and [η5-2-(9 anthracenyl)indenyl]tricarbonylrhenium, **29**.

As depicted in Figure 13, in 29, oscillation of the (indenyl)Re(CO)3 moiety across the mirror plane containing the anthracenyl framework, which equilibrates each of the three pairs of indenyl protons but maintains the inequivalence of the benzo rings of the anthracene, has a barrier too low to be accessed by V-T NMR even at −80 ◦C. In contrast, a window-wiper type of motion that equilibrates the terminal rings of the anthracene has to surmount a 48 kJ mol−<sup>1</sup> barrier that arises because of the build-up of steric interactions as the indenyl and anthracenyl fragments approach coplanarity. Complete 360◦ rotation to generate effective time-averaged *C*2v symmetry requires that both processes are operative [20].

The analogous η6-chromium and η5-rhenium complexes of 9-(3-indenyl)anthracene, 31 and 32, respectively, have also been prepared and structurally characterised (Figure 14). In these molecules, the interplanar indenyl-anthracenyl dihedral angles are 75.5◦ and 56.5◦, respectively. As previously discussed, in the free ligand steric interactions between the peri-hydrogens, H(1) and H(8), of the anthracene skeleton and the indenyl substituent give rise to a rotational barrier of ca. 105 kJ mol−1. Following the pattern seen with [η5-2-(9-anthracenyl)indenyl]tricarbonylrhenium, 29, the barrier in 32 was reduced to 90 kJ mol−1, while that for the chromium complex, 31, was somewhat higher at ~96 kJ mol−1. Furthermore, as shown in Scheme 11, deprotonation of 31 initiated the η<sup>6</sup> to η<sup>5</sup> haptotropic shift, forming anion 33, in which the rotational barrier had once again fallen to 90 kJ mol<sup>−</sup>1.

**Figure 13.** Exchange behaviour in the indenyl-anthracene rhenium complex **29**. (**i**) Low-energy (Cs symmetric) up/down oscillation of the indenyl-Re(CO)3 group relative to the anthracene ring plane; (**ii**) higher-energy process whereby side-to-side rotation of the indenyl-Re(CO)3 moiety about the C(9)-indenyl linkage equilibrates the terminal benzo rings.

**Figure 14.** Molecular structures of [η6-3-(9-anthracenyl)indene]tricarbonylchromium, **31**, and [η5-3-(9 anthracenyl)indenyl]tricarbonylrhenium, **32**.

#### *2.5. Organometallic Derivatives of Indenyl Triptycenes*

Attempts to coordinate M(CO)3 tripods, where M = Cr, Mn or Re, to indenyl triptycenes led to a number of different products (Scheme 12). Dealing first with 9-(3-indenyl)triptycene, 8, reaction with Cr(CO)6 gave only a single isomer, 34, in which the tricarbonylchromium fragment is sited on a benzo ring of the triptycene (Figure 15), rather than on the six-membered ring of the indenyl substituent [10]. An alternative approach, addition of benzyne to 31, in which the Cr(CO)3 moiety is already in place, was unsuccessful. However, reaction of 8 with Re2(CO)10 gave a low yield of the η5-Re(CO)3 complex, 35 [20]. We note in particular that the indenyl ring in 35 is oriented such that the tricarbonylrhenium tripod is aligned with a valley between two blades (Figure 15).

The result of coordinating metal carbonyl tripods to 9-(2-indenyl)triptycene, 9, was more straightforward, and ultimately more satisfying [21]. As depicted in Scheme 12, the reaction with Cr(CO)6 furnishes two products, the target molecule, 36, in which the Cr(CO)3 unit is coordinated to the six-membered ring of the indene (Figure 16), and also its triptycyl blade complexed isomer, 37. Concomitantly, reaction with Re2(CO)10 delivered the η5-Re(CO)3 complex 38. The analogous (and isostructural) η5-Mn(CO)3 complex, 39, was also prepared, by reaction of the deprotonated ligand with BrMn(CO)5.

**Scheme 12.** Cr and Re complexes of 9-(3-indenyl)triptycene and 9-(2-indenyl)triptycene.

**Figure 15.** Structures of [η6-9-(3-indenyl)-1,2,3,4,4a,9a-triptycene]tricarbonyl-chromium, **34**, and [η5-3-(9-triptycyl)indenyl]tricarbonylrhenium, **35**.

**Figure 16.** Structures of [η6-2-(9-triptycyl)indene]tricarbonylchromium, **36**, and [η5-2-(9 triptycyl)indenyl]tricarbonylrhenium, **38**.

A 13C NMR study on the η5-rhenium complex 38 revealed that, at room temperature, the triptycyl resonances are each split into a 2:1 pattern, clearly indicating that paddlewheel rotation was slow on the NMR time-scale. However, upon raising the temperature, the gradual onset of line broadening, together with computer simulation of the spectra, yielded a barrier of 84 <sup>±</sup> 2 kJ mol−<sup>1</sup> for equilibration of the three benzo blades; the manganese analogue, 39, behaved similarly. In contrast, in the η6-chromium complex, 36, peak decoalescence even at low temperatures was never apparent, revealing that paddlewheel rotation continued essentially unhindered.

The crucial experiment whereby deprotonation of 36 to form its η5-haptotropomer, 40, once again brought about a 2:1 splitting of the triptycyl blade NMR resonances is depicted in Scheme 13. The space-filling representation (Figure 17) of the η5-Mn(CO)3 complex, 39, which is an ideal structural model for the isoelectronic anion, 40, illustrates unequivocally how the bulky organometallic group obtrudes into a valley between two benzo blades, thus hindering paddlewheel rotation. One can only conclude that shuttling of the Cr(CO)3 moiety from the six-membered to the five-membered ring of the 2-indenyl component, brought about by deprotonation, represents a pH-dependent organometallic molecular brake [21].

**Scheme 13.** Deprotonation of **<sup>36</sup>** brings about an <sup>η</sup><sup>6</sup> <sup>→</sup> <sup>η</sup><sup>5</sup> haptotropic shift, forming **<sup>40</sup>**, in which paddlewheel rotation (dashed arrow) is dramatically slowed on the NMR time-scale.

**Figure 17.** Space-fill view of [η5-2-(9-triptycyl)indenyl]tricarbonylmanganese, **39**, showing how a carbonyl ligand is positioned directly between two blades of the triptycene.

#### **3. Ferrocenyl Anthracenes and Triptycenes**

#### *3.1. Mono- and di-Ferrocenyl Anthracenes*

The demonstration of an organometallic molecular brake involving an η<sup>6</sup> Ɂ η<sup>5</sup> haptotropic shift driven by a deprotonation/protonation sequence prompted us to consider the possibility of using an electrochemically-driven redox approach. To this end, we wished to incorporate archetypal bulky redox-active species, viz. one or more ferrocenyl units. The reported synthesis of 9-ferrocenylanthracene, 41, by Butler, utilising the Negishi-type cross-coupling reaction of chlorozincioferrocene with 9-bromoanthracene using (dppf)PdCl2 as the catalyst, led to yields of 30–35% [22]. Previous attempts to bring about a Suzuki cross-coupling were unsuccessful unless a large excess of ferrocenylboronic acid was used along with an almost stoichiometric amount of catalyst [23]. However, when tetrabutylammonium hydroxide in 1,4-dioxane was used as the base and (dppf)PdCl2 as the catalyst, yields up to 90% were achievable (Scheme 14) [24].

**Scheme 14.** Syntheses of mono- and di-ferrocenyl anthracenes and triptycenes. Reagents and conditions: (**i**) ferrocenylboronic acid, Bu4NOH, (dppf)PdCl2, 1,4-dioxane, 24 h, 120 ◦C; (**ii**) *O*-BrC6H4F, BuLi, toluene, −5 ◦C.

As shown in Figure 18, the ferrocenyl substituents in 9-ferrocenylanthracene, 41, and 9,10 diferrocenylanthracene, 42, each make dihedral angles of 45◦ with the plane of the anthracene; in 42, the ferrocenyls are rotated 89◦ from each other, thus engendering *C*<sup>2</sup> symmetry. However, it is evident from our detailed V-T experiments that 42 can exist in two forms, and at 193 K these atropisomers are indeed detectable by NMR in a 38:62 *syn* to *anti* ratio.

**Figure 18.** Structures of 9-ferrocenylanthracene, **41**, and 9,10-diferrocenylanthracene, **42**.

The dynamic behaviour of 9-ferrocenylanthracene, 41, paralleled that of [η5-2-(9 anthracenyl)indenyl]tricarbonylrhenium, 29, whereby, entirely analogous to that depicted in Figure 12, the system racemises, via a low-energy process, by oscillation of the ferrocenyl group about the mirror plane containing the anthracene, thereby exhibiting dynamic *C*s symmetry. The second, higher-energy, process has a barrier of 44 <sup>±</sup> 2 kJ mol<sup>−</sup>1, and together these allow the ferrocenyl moiety to access both faces of the anthracene and both terminal benzo rings; thus, at room temperature, the molecule exhibits effective *C*2v symmetry on the NMR time-scale.

In the structure of 42, depicted in Figure 18, the two ferrocenyl fragments are positioned on the same face of the anthracene, but, because there is a substantial barrier (ca. 45 kJ mol<sup>−</sup>1) towards either of them becoming coplanar with the central anthracene ring, this actually represents an anti-isomer with dynamic *C*2h symmetry.

#### *3.2. Mono- and di-Ferrocenyl Triptycenes*

Addition of benzyne to 41 and 42 yields the corresponding ferrocenyl triptycenes, 43 and 44, respectively (Figure 19). On a 500 MHz spectrometer, even at room temperature, the signals of the three benzo blades of 43 are split into a 2:1 pattern, indicating slowed paddlewheel rotation on the NMR time-scale; the barrier was evaluated as 69 <sup>±</sup> 2 kJ mol−1. As with its anthracene precursor, 9,10-diferrocenyltriptycene, 44, exists as two atropisomers, whereby the two sandwich moieties are aligned in the same valley (eclipsed, *C*2v, *meso*) or in different valleys (gauche, *C*2, *racemic*), and are readily distinguishable by NMR spectroscopy [24].

**Figure 19.** Structures of 9-ferrocenyltriptycene, **43**, and *rac*-9,10-diferrocenyltriptycene, **44**.

An Exchange Spectroscopy (EXSY) study of the dynamic behaviour of 9,10-diferrocenyltriptycene revealed a number of interesting features, in particular the nature of the exchange mechanism linking the rotamers of 44. Thus, interconversion of the two mirror forms of the *rac* isomer proceeds in a stepwise manner via the *meso* structure; likewise, a 120◦ rotation of a single ferrocenyl in the *meso* isomer generates a *rac* structure, and these interconversions are readily detectable in the EXSY experiment when the mixing time is short (50 ms). However, with longer mixing times (300 ms), second generation cross-peaks become evident, thereby revealing exchange between sites within the *rac* or *meso* isomers. This can only occur via the other rotamer, that is, it must follow the sequence *meso* → *rac* → *meso* or *rac* → *meso* → *rac* [24].

We note that the assignment of 1H NMR resonances in the vicinity of a ferrocenyl group is greatly facilitated by its extraordinarily large diamagnetic anisotropy [25]. Typically, in ferrocenyl triptycenes, a proton lying directly above a cyclopentadienyl ring is shielded by ~1.2 ppm relative to its normal aromatic resonance position. In contrast, protons lying near the horizontal plane containing the iron atom are deshielded by ~1.2 ppm, as exemplified by the data for the *C*1-symmetric rotamer of 2,3-dimethyl-9-ferrocenyltriptycene whose structure is shown in Figure 20.

**Figure 20.** 1H NMR chemical shifts (ppm) in 2,3-dimethyl-9-ferrocenyltriptycene.

This work has been extended to a situation where the triptycene itself is dissymmetric by virtue of the presence of two bulky tert-butyl substituents [26]. Suzuki cross-coupling of 9,10-dibromo-2,6 di-tert-butylanthracene with ferrocenylboronic acid catalysed by (dppf)PdCl2 in dioxane gave 45 (Figure 21) in 93% yield, and subsequent Diels-Alder addition of benzyne delivered the required 2,6-di-tert-butyl-9,10-diferrocenyltriptycene, 46 (Scheme 14).

**Figure 21.** Molecular structure of 2,6-di-tert-butyl-9,10-diferrocenylanthracene, **45**.

Since each ferrocenyl unit in the triptycene 46 can adopt one of three different positions, one can envisage nine atropisomers: three eclipsed structures and six gauche conformers. However, symmetry considerations lower this number to six different rotamers, three of which are doubly degenerate. Stepwise interconversions of these rotamers do not all have identical barriers, since manoeuvring around a bulky tert-butyl substituent is more sterically demanding than the traversal past a C-H linkage. Nevertheless, as noted above, the diamagnetic anisotropy of the ferrocenyl fragments dramatically enhances the separation between aromatic proton resonances such that, at 600 MHz, the problem becomes resolvable, and all six rotamers can be unequivocally identified. Indeed, in Figure 22, are shown the exchange pathways between the six rotamers; in some cases, direct interconversion between rotamers is possible (such as A Ɂ E), in others a two-step process (e.g., B Ɂ C) is mandatory. Note that the lowest energy route from A to D goes in two steps via E.

**Figure 22.** Possible one-step interconversion pathways between conformers of 2,6-di-tert-butyl-9,10-diferrocenyltriptycene, 46. Green and red arrows indicate high and low barriers, respectively. Structures shown in blue are C2-symmetric, and those possessing only C1 symmetry are in black [26].

#### *3.3. Cycloaddition Reactions of Mono- and di-Ferrocenyl Anthracenes*

Cycloaddition reactions of anthracenes with a wide range of dienophiles have been intensely studied ever since the original work of Diels and Alder [27], and the relationship between electronic and steric effects is complex. Typically, benzyne adds to 9,10-dimethylanthracene to form exclusively the electronically favoured 9,10-dimethyltriptycene, thus generating three aromatic systems. In contrast, with 9,10-diphenylanthracene, benzyne adds to the 1,4 rather than the 9,10-positions by a factor of 93:7 [28].

Since palladium cross-coupling procedures had provided us with sufficient quantities of monoand di-ferrocenyl anthracenes, 41 and 42, to enable further investigation of their reactivity, we undertook a comprehensive study of their cycloaddition chemistry, in particular their reactions with benzynes, alkynes, maleimides and benzoquinone. As shown in Scheme 15, benzyne, dimethylbenzyne and fluorobenzyne add across C(9) and C(10) to form triptycenes, whereas trifluoromethylbenzyne adds in a 1,4-fashion to give the corresponding tetracene, and tetrafluorobenzyne adds both ways. Likewise, DMAD, N-methyl- and N-phenyl-maleimides and benzophenone all react to yield the appropriate barrelene [29,30]. The more sterically hindered 9,10-diferrocenylanthracene also undergoes 9,10-additions with benzynes bearing small substituents, but trifluoromethylbenzyne, DMAD and the maleimides add predominantly at the C(1) and C(4) positions (Scheme 16).

**Scheme 15.** Selected cycloaddition reactions to 9-ferrocenylanthracene.

**Scheme 16.** Selected cycloaddition reactions to 9,10-diferrocenylanthracene.

#### **4. Syntheses and Dynamic Behaviour of Biindenyls**

#### *4.1. Cross-Coupling of 2-Phenyl- and 2-Methyl Indenes*

The CuCl2-mediated reaction of either 2-phenyl or 2-methylindene generates *racemic* and *meso* biindenyls (Scheme 17) in almost equal amounts, implying a radical process. Surprisingly perhaps, simply allowing their lithio salts to react slowly with oxygen yields only *racemic* products; this has been interpreted in terms of a nucleophilic attack by one anion on the peroxide of its partner, but the mechanism remains speculative [31].

**Scheme 17.** Formation of *racemic* and *meso* bi-indenyls **42**–**45**.

The structures of molecules 47 through 50 are shown in Figures 23 and 24 and, in all cases, the indenyl fragments adopt a gauche orientation [19,31]. It is interesting to note that the *racemic* products always maintain their *C*<sup>2</sup> symmetry whatever the rotation angle between the two indenyl units. In contrast, *meso* isomers lose all their symmetry elements unless the central H-C-C-H dihedral angle is 0◦ or 180◦, thereby giving rise either to a mirror plane (*C*s) or an inversion centre (*C*i), respectively. The rotational barriers about the axis linking the indenyls are ca. 85 and 65 kJ mol−<sup>1</sup> in the phenyl and methyl cases, respectively [19].

**Figure 23.** Molecular structures of *meso*- and *rac*-2,2 -diphenyl-1,1 -biindenyl, **47** and **48**.

**Figure 24.** Molecular structures of *meso*- and *rac*-2,2 -dimethyl-1,1 -biindenyl, **49** and **50**.
