A Brief History of Photoactive Interlocked Systems Assembled by Transition Metal Template Synthesis

Abstract

More than three decades of research efforts have yielded powerful methodologies based on transition metal template-directed syntheses for the assembly of a huge number of interlocked systems, molecular knots, machines and synthesizers. Such template techniques have been applied in the preparation of mechanically linked electron donor–acceptor artificial photosynthetic models. Consequently, synthetic challenging photoactive rotaxanes and catenanes have been reported, in which the chromophores are not covalently linked but are still associated with undergoing sequential energy (EnT) and electron transfer (ET) processes upon photoexcitation. Many interlocked photosynthetic models produce highly energetic, but still long-living charge separated states (CSS). The present work describes in a historical perspective some key advances in the field of photoactive interlocked systems assembled by transition metal template techniques, which illustrate the usefulness of rotaxanes and catenanes as molecular scaffolds to organize electron donor–acceptor groups. The effects of molecular dynamics, molecular topology, as well as the role of the transition metal ion used as template species, on the thermodynamic and kinetic parameters of the photoinduced energy and electron transfer processes in the interlocked systems are also discussed.

1. Introduction

The mechanical bond is defined as an entanglement in space between two or more component parts, such that they cannot be separated without breaking or distorting chemical bonds between atoms. One of the most interesting features of interlocked systems is that the mechanical bond restricts the degrees of freedom of the molecular components when compared to a mixture of the molecules in solution. Therefore, molecular dynamics can be better controlled in interlocked systems, very often through external stimuli, to yield relatively simpler dynamic systems that show emergent properties. Accordingly, the exploration of the dynamic properties in molecular interlocked systems through control of the relative motions of the mechanically linked molecular components has led to major advances in several scientific endeavors as well as to the development of the most popular and Nobel Laureate-producing research field of molecular machines [1,2,3,4,5,6,7,8].

The impressive advancements in the field of mechanically linked molecules we are now witnessing have been possible due to the development of reliable synthetic protocols to assemble interlocked systems in the early 1980s. The formation of mechanical bonds is thermodynamically prohibitive due to the large entropic penalty involved in the self-assembly of the molecular components dissolved in solution into the relatively higher-organized interlocked systems. To overcome such entropic costs, the formation of mechanical bonds requires template synthetic techniques, which are based on molecular recognition through formation of attractive secondary interactions between the molecular components in solution. Secondary interactions yield enthalpic gains, which can eventually overcome the entropic hurdles to drive the self-assembly of the molecular components into the specific molecular architectures that favor formation of mechanical bonds. Accordingly, hydrogen bonding, π-π stacking, hydrophobic interactions, ion- and radical-pairing have been largely explored as template routes for molecular recognition and development of several efficient and elegant synthetic template techniques to afford interlocked systems. The prime examples of interlocked systems are rotaxanes (a dumbbell shaped molecular component in which a molecular ring is trapped) and catenanes (two or more interlocked rings) [1,9,10,11,12,13,14,15].

Among the template synthetic techniques published in the literature, the so-called transition metal template methodologies, which are based on the formation of coordinative bonds between functionalized ligands and transition metal ions as template species, have proven to be a powerful tactic to prepare molecular racks, ladders, squares, grids, cylinders, molecular boxes, knots and double-, triple-, and circular helicates, as demonstrated by Saalfrank [16,17], Lehn [9] and others [18]. In the field of interlocked molecules, this approach is best illustrated using Sauvage’s pioneering and elegant 1,10-phenanthroline-copper(I) metal template strategy as example for the synthesis of catenanes (Figure 1) [19]. In this approach, the Cu(I) metal template ion gathers and positions the 2,9-diphenyl-1,10-phenanthroline bidentate ligands (phen) to form an entwined [Cu(phen)₂]⁺ complex through formation of coordinative bonds. As elegantly demonstrated by Schmittel [20,21], the formation of secondary π-π attractive interactions between the phenyl rings and the phen moieties provide the entwined [Cu(phen)₂]⁺ complex with extra stability and thereby are critical for the success of Sauvage’s catenane metal template synthesis. Accordingly, the distorted tetrahedral geometry of the resulting [Cu(phen)₂]⁺ complex forces the phen ligands to physically overlap, which in turn creates the cross-over point that directs the hydroxyl groups on the phen ligands to point in opposite directions. Furthermore, the proximity of the hydroxyl groups in the entwined [Cu(phen)₂]⁺ complex favors the intramolecular macrocyclization Williamson reactions to afford catenanes, rather than deleterious intermolecular process that would yield oligomers and/or higher cyclic species. The removal of the Cu(I) template ion at the final step yields the target catenane [5,8,19]. As we shall see later, the strategy is easily adapted to the synthesis of rotaxanes. Later, Leigh and collaborators published a series of influential studies, which include other transition metal ions as template species, which have established the guidelines for diverse synthetic strategies and tactics to efficiently assemble rotaxanes, catenanes, molecular knots and higher order links via the formation of complexes with several geometries [22].

The elegant developments in the template synthesis of interlocked molecules and further exploration of the interesting dynamic properties of the mechanically linked systems that followed Sauvage’s pioneering works attracted the attention of the scientific community working on artificial photosynthetic models [23,24,25,26,27,28,29,30,31]. The possibility to introduce mechanical bonds into electron donor–acceptor systems (D-A) opened the door to design structurally more complex D-A models in which the effects of molecular dynamics and molecular topology on the thermodynamic and kinetics of forward (ET) and back electron transfer (BET) reactions could be investigated. Many mechanically linked D-A systems assembled via template methodologies based on the formation of several secondary interactions have been published in the literature [32,33,34,35,36,37,38,39,40,41,42,43,44,45,46,47,48,49,50,51,52]. However, the transition metal template technique has become the most popular one as it offers an extra advantage over the other template techniques, namely the entwined metal complexes necessary to assemble the interlocked systems can also function, in most cases, as electronic relays [29,53,54,55].

In this context, we describe herein a brief overview of mechanically linked artificial photosynthetic models assembled by transition metal template techniques. There are a significant number of D-A interlocked photoactive systems reported in the literature that have been prepared from template synthetic methodologies based on the formation of secondary interactions other than coordinative bonds. Those works are beyond the scope of this account. The text is organized in a historical perspective, briefly describing the synthetic strategies to assemble the photoactive interlocked systems, followed by a discussion of their photophysical properties. The present text is by no means a review of the field, but rather a short account with illustrative examples of key advances in the field. For extensive reviews, we direct the reader to several excellent works that have been published in the literature [1,24,31,36,56,57,58].

2. Interlocked Photosynthetic Models Decorated with Porphyrins as Electron Donors and Acceptors

Sauvage’s group was the first to realize the potential of investigating rotaxanes and catenanes as interactive scaffolds to organize electron donor–acceptor moieties in artificial photosynthetic models. In their pioneering works [59,60,61,62,63,64,65,66,67,68,69], a large family of photoactive rotaxanes was synthesized and their photophysical properties were investigated. To illustrate Sauvage’s pioneering works, the synthesis, dynamic and photophysical properties of rotaxane 1 will be discussed here (Figure 2) [59,60,61]. Rotaxane 1 was designed to have a cationic Au(III)porphyrinate group (AuP) as the electron acceptor, which was covalently attached to the ring component of the rotaxane, while the two Zn(II)porphyrinate stoppers (to prevent dissociation of the rotaxane upon removal of the Cu(I) template ion) were the electron donors. The synthesis of rotaxane 1 was achieved from an adaptation of the original Cu(I)-metal template synthetic strategy (Figure 1) and relied on the ‘‘gathering and threading’’ effects promoted by the Cu(I) ion template. Accordingly, a dialdehyde phen-based stringlike fragment was threaded through a previously prepared phen-containing ring with the appended AuP group to quantitatively form the entwined [Cu(phen)₂]⁺ complex that is also called pseudorotaxane. Finally, the two ZnP units were prepared simultaneously by reaction between the aldehyde functionalities on the thread component of the pseudorotaxane and dipyrromethane under acidic conditions (17% yield), followed by metalation of the free-base porphyrins with Zn(OAc)₂, to afford rotaxane 1 (82% yield).

Photophysical investigation of rotaxane 1 revealed that excitation of the ZnP chromophores yielded the corresponding singlet state (¹ZnP*), which partially decayed by ET (70% yield, k = 1.6 × 10¹⁰ s⁻¹) to the AuP acceptor and 30% involving energy transfer (EnT) to the [Cu(phen)₂]⁺ complex (k = 3.3 × 10⁹ s⁻¹). The resulting ZnP^•+–[Cu(phen)₂]⁺–AuP^• charge separated state (CSS) afforded from the ET was found to biexponentially decay with lifetimes within 10–40 ns via BET. The biexponential decay of the CSS informed that rotaxane 1 was conformationally flexible, which allowed the photoactive subunits to have distinct electronic couplings. Exclusive excitation of the AuP moiety in 1 yielded the ¹AuP*, which was reductively quenched by the ZnP moieties via ET mediated by the [Cu(phen)₂]⁺ complex to afford the same ZnP^•+–[Cu(phen)₂]⁺–AuP^• CSS [60,61,62,63,64,65,66,67,68].

Technically speaking, rotaxane 1 does not bear a mechanical bond, as the rigid tetrahedral metal chelate prevents the rotaxane components from undergoing the relatively long-range molecular motions about each other. Therefore, the authors removed the Cu(I) template ion by treating 1 with an excess of KCN to investigate the possible rearrangements of the chromophores in the Cu-free rotaxane. From a detailed spectroscopic investigation by NMR techniques, it was demonstrated that upon demetallation of 1, the ring component circumrotated about the thread to position the AuP group between the tweezers-like configuration of the ZnP stoppers. This dynamic process was driven by attractive π-π interactions between the chromophores, which yielded the most stable ZnP-AuP-ZnP triple decker molecular configuration, schematically shown in Figure 2 [62]. However, due to the proximity of the chromophores in the Cu-free rotaxane, many intermediates produced upon photoexcitation had too short lifetimes to be investigated in the timeframe of the available spectrometers (20 ps), thus precluding a detailed investigation of the photophysical properties of the demetallated rotaxanes. Despite those limitations, Sauvage and coworkers have demonstrated for the first time that the inherent dynamic processes brought by mechanical bonds could be used to rearrange the photoactive subunits in rotaxanes to completely change the kinetics of ET and BET. The same group also reported a series of similar multirotaxanes and the parent catenanes, whose syntheses and photophysical properties have already been reviewed [57,69].

3. Interlocked Photosynthetic Models Decorated with Porphyrins as Electron Donors and Fullerenes as Acceptors

The pioneering works by Sauvage and collaborators brought much insight into the effects of molecular topology on the thermodynamics and kinetics of photo-induced processes. However, in addition to the already mentioned ultrafast ET and BET processes of the ZnP-AuP-based rotaxanes and catenanes, another limitation found in those pioneering works was the overlapping of the spectroscopic signals of the many intermediates formed upon excitation. Accordingly, a complete determination of the kinetic parameters for the expected photophysical decays was impossible. The solution to this problem was to replace the AuP acceptor with [60]fullerene (C₆₀) in the interlocked molecular architectures. This idea came from the vast experimental evidence accumulated from the investigations of covalently linked D-A photoredox arrays published in the literature [70,71,72,73,74,75,76,77,78]. Such works clearly demonstrated that C₆₀ was a superior electron acceptor than AuP and other chromophores commonly used in artificial photosynthetic models at that time, such as quinones [79]. Those studies confirmed that the C₆₀ rigid structure associated with its poor solvation yielded small values of reorganization energy (λ), which shifted the wasteful BET processes into the inverted region of the Marcus parabolic relationship between free energy change of the ET processes and λ [80]. Accordingly, long-lived CSSs in many covalently linked artificial photosynthetic models containing C₆₀ as acceptors were reported. Furthermore, the transient absorption spectroscopic signature of the reduced fullerene radical anion (C₆₀^•–) appears at about λ_max ≈ 1000 nm, which is usually a clean region of the absorption spectrum, thereby solving the signal overlapping issues in previous photophysical investigations [70,71].

Sauvage in collaboration with Diederich and Nierengarten reported the first rotaxane containing C₆₀ as the electron acceptor. In their design, the rotaxane was assembled via the Cu(I) metal template technique and the C₆₀ groups functioned as stoppers in the interlocked photoactive model (Figure 3a) [81]. The synthetic strategy used to prepare target bis-C₆₀-rotaxane 2, which was isolated in 15% yield, was based on Hay oxidative alkyne–alkyne coupling to introduce the fullerene groups into the [Cu(phen)₂]⁺ pseudorotaxane. The non-interlocked thread compound shown in Figure 3a was also isolated from the crude product, thus informing that the central [Cu(phen)₂]⁺ complex in the pseudorotaxane partially dissociated during the Hay-stoppering reactions. Investigation of the redox properties of rotaxane 2 revealed a significant anodic shift (~0.2 V) for the reversible oxidation of the [Cu(phen)₂]⁺ complex (0.87 eV vs. SCE) in comparison with a reference compound (0.68 eV vs. SCE) lacking the fullerene groups. Reduction (irreversible) of the C₆₀ component appeared in the typical range (0.6 eV vs. SCE), leading the authors to speculate that the strong electron-withdrawing effect of the fullerenes might substantially destabilize the highest oxidation state of the [Cu(phen)₂]⁺ complex [82].

Time-resolved spectroscopic investigation revealed that excitation of rotaxane 2 at 355 or 532 nm simultaneously yielded the metal-to-ligand charge transfer (MLCT*) in the [Cu(phen)₂]⁺ complex (40% yield) and the ¹C₆₀^* (60% yield) excited states (step 1, Figure 3b). The authors proposed that the ¹C₆₀^* decayed via energy transfer (EnT) to the [Cu(phen)₂]⁺ complex (62% yield, k = 1.6 × 10⁹ s⁻¹, step 2) in competition with intersystem crossing (28% yield, k = 6.2 × 10⁸ s⁻¹ to generate ³C₆₀^*, step 3). From the electrochemical measurements, both MLCT* and ¹C₆₀^* should have decayed via ET to form the [Cu(phen)₂]²⁺–C₆₀^•− CSS (steps 4 and 5). A weak transient absorption signal at 740 nm with a lifetime of 1.7 μs was observed, which could have been attributed to formation of the [Cu(phen)₂]²⁺–C₆₀^•– CSS. However, the unusually long lifetime of this transient species along with the impossibility of observing the spectroscopic signature of the C₆₀^•− radical anion around λ = 1000 nm due to instrumental limitations led the authors to conservatively conclude that the [Cu(phen)₂]²⁺-C₆₀^•− CSS was probably formed in rotaxane 2, but BET (step 8) was too fast to allow unambiguous detection of the expected [Cu(phen)₂]²⁺-C₆₀^•− CSS [82].

Based on those previous works, Schuster’s and Guldi’s groups initiated a research program to further improve the synthetic methodologies to prepare Sauvage-type interlocked systems containing C₆₀ as electron acceptor, while overcoming the instrumental limitations that had prevented the complete elucidation of photo-induced processes in the rotaxanes and catenanes. In their initial efforts [83], Schuster and colleagues elaborated elegant synthetic strategies to afford a new family of [Cu(phen)₂]⁺-based rotaxanes, in which the ZnP electron donors were positioned at the end of the phen-thread component, while the C₆₀ acceptor was covalently attached to the phen-macrocycle (Figure 4) using classical fullerene chemistry [73].

Schuster’s synthetic strategies were conceived to minimize the usual C₆₀ solubility issues and the inherent kinetic lability of the coordinative bonds that held together the [Cu(phen)₂]⁺ complex. Accordingly, the new family of photoactive rotaxanes were prepared following a stepwise approach. For illustrative purposes, the synthesis of rotaxane 3 will be described (Figure 5). Starting with phen-macrocycle 6, the malonate synthon reacted smoothly with C₆₀ under Bingel–Hirsch conditions [73] to yield compound 7, which was soluble in most organic solvents. The mono-ZnP-stoppered thread 10 was prepared from tetraarylporphyrin carboxylic acid 8 and phen-thread 9 via esterification reaction using dicyclohexylcarbodiimide (DCC) as coupling agent and 4-dimethylaminopyridine (DMAP) as catalyst. The ‘‘threading’’ reaction of the mono-ZnP-stoppered phen-stringlike fragment 10 through macrocycle 7 was accomplished using the Cu(I) ion as the template species to yield the [Cu(phen)₂]⁺–C₆₀ pseudorotaxane precursor 11, which was known to be less prone to dissociation [17], thereby yielding rotaxanes in larger yields. In the final step, a second DCC/DMAP ester-coupling reaction between pseudorotaxane 11 and tetraarylporphyrin carboxylic acid 8 afforded target rotaxane 3 in about 40% yield.

An elegant series of electrochemical, time-resolved emission and transient absorption experiments was then carried out on the new family of rotaxanes and related model compounds by Echegoyen’s and Guldi’s groups. Such detailed investigation enabled the authors to unambiguously assign the specific roles of each entity in the rotaxanes, thereby allowing the determination of the kinetics of the photoinduced processes in the interlocked molecules (Figure 6). Exclusive excitation of the ZnP subunits at 420 nm yielded the ¹ZnP* excited state (step 1), which was moderately quenched (the fluorescence lifetimes of the ¹ZnP* were 3.2 ns in the reference compound and 1 ns in the rotaxanes). Furthermore, the emission quenching was independent of the solvent polarity and present in rotaxane model compounds lacking the fullerene moiety, which informed the authors that the quenching did not involve direct interaction of ¹ZnP* and C₆₀. Accordingly, the authors attributed the quenching to EnT from the ¹ZnP* to the [Cu(phen)₂]⁺ subunit to yield the corresponding MLCT* excited state (step 2). By monitoring the transition absorption signal of the C₆₀^•− at around λ_max = 1000 nm in the rotaxanes, which obeyed a biexponential rate law, along with a careful transient absorption investigation of several model compounds, the authors determined that the MLCT* manifold was oxidatively quenched by the C₆₀ via ET to yield the intermediate ZnP–[Cu(phen)₂]²⁺–C₆₀^•− CSS, which partially yielded the ground state via BET (steps 3 and 5, respectively). However, detection of the strong signature transient absorption of the ZnP^•+ centered at λ_max = 680 nm, whose lifetime matched the long component decay observed for the biexponential rate law of the fingerprint absorption of the C₆₀ at λ_max = 1040 nm, provided clear cut evidence for an ET reaction from the ZnP stoppers to the oxidized [Cu(phen)₂]²⁺ complex to afford the final ZnP^•+–[Cu(phen)₂]⁺–C₆₀^•− CSS (step 4). The lifetimes of the final CSSs in the rotaxanes were in the range of 0.4–1.4 μs (step 6). Such relatively long lifetimes of the final CSSs allowed the authors to conclude that BET in the rotaxanes occurred in the Marcus inverted region. The main photophysical decay of Schuster’s and Guldi’s rotaxanes are summarized in the energy level diagram depicted in Figure 6.

Based on those promising findings, Schuster, Guldi and collaborators reported a second set of rotaxanes [84], in which the positions of the ZnP and C₆₀ groups were reversed. Accordingly, in this new set of rotaxanes, the ZnP moiety was covalently attached to the macrocycle component of the rotaxane, while the fullerenes acted as stoppers on the thread. Following a similar stepwise strategy, rotaxanes 12–14 were assembled (Figure 7) and their interlocked structures confirmed by NMR spectroscopic investigation. The overall sequence of photophysical events in the fullerene-stoppered rotaxanes was identical to that shown in Figure 6, thereby confirming the assignments of the role of each photoactive subunit in the rotaxanes. However, the lifetimes of the final ZnP^•+–[Cu(phen)₂]⁺–C₆₀^•− CSSs were much longer for the second set of rotaxanes, particularly that observed for rotaxane 12, which was 32 μs in THF and was among the longest ever reported for the CSS state of a simple porphyrin–fullerene dyads at the time. The authors suggested that such distinctly lifetimes observed for their D-A rotaxanes could be due to a variety of factors, including D-A distance, the chemical nature of the linker between the chromophores, and effects of molecular topology on the kinetics of the photoinduced processes [85].

4. Effects of Molecular Topology on the Kinetics of Photoinduced Processes in Interlocked Molecules Decorated with Porphyrins as Electron Donors and Fullerenes as Acceptors

From the achievements described in the previous section, along with the increasing amount of experimental evidence reported in the literature for D-A rotaxanes assembled by different template techniques, it became clear that small structural modifications, such as variation of the size and shape of the ring and thread components, as well as the type of chemical linkages that connect the chromophores to the interlocked architectures, significantly affected the kinetics of the photoinduced processes. In other words, investigation of how molecular topology affected the ET and BET processes was of paramount importance in the field. In chemistry, molecular topology generally refers to molecules whose representations based on atoms and bonds are nonplanar. In chemical topology, a molecule can be distorted as much as one wants, but its topological properties are not modified as long as no cleavage of a chemical bond occurs. According to this definition, rotaxanes and catenanes are topologically different. For example, in a [2]catenane, the two mechanically linked rings are topologically distinct from the two molecular rings that are not interlocked. Therefore, catenanes are nontrivial topological systems. Conversely, a [2]rotaxane is topologically identical to its component parts as the ring can be separated from the thread by stretching or compressing the appropriate chemical bonds [1,4].

In this context, the best way to investigate the effects of molecular topology on the kinetics of the photoinduced processes was to synthesize rotaxanes and catenanes with identical molecular parameters and decorated with ZnP and C₆₀ chromophores. Although Schuster and coworkers demonstrated that ester-coupling reactions were suitable synthetic methodologies for the preparation of ZnP–C₆₀ rotaxanes, they failed in the preparation of the analogous catenanes. To overcome the synthetic challenges in the preparation of photoactive catenanes, Megiatto and Schuster developed an innovative synthetic strategy (Figure 8) based on “click reactions” [86,87] as the final step that transforms the [Cu(phen)₂]⁺-based pseudorotaxane 17 into target rotaxane 18 (via stoppering reactions using 15) and catenane 19 (via macrocyclization reactions using 16). For the success of this strategy, the final “click reactions” had to occur fast in order to favorably compete with the kinetic labile coordinative bonds that hold the components in the entwined configuration in pseudorotaxane 17. Otherwise, a detrimental unthreading process via dissociation of the [Cu(phen)₂]⁺ complex would take place and the components would react in a random rather than in a templated fashion.

The hurdles of the proposed “click” synthetic strategy became obvious in the initial efforts. Classical tactics to improve the kinetics of a chemical reaction, such as increasing the concentration of the reactants and/or adding an excess of one of the reactants, worked well for the synthesis of rotaxane 18, but completely failed when trying to prepare the analogous catenane 19, as intractable product mixtures containing oligomers and polymers were afforded under those conditions. On the other hand, the typical homeopathic reactant dilution necessary for efficient macrocylization reactions yielded target catenane 19. However, such high-dilution conditions, when applied to the preparation of rotaxane 18, led to very poor yields, most likely because the lower ‘‘click’’ reaction rates allowed dissociation of pseudorotaxane 17 in the route to rotaxane 18.

To overcome those diverging reaction conditions and afford both rotaxanes and catenanes from the same experimental conditions, a new “click” catalyst brew was developed to speed up the final stoppering and macrocyclization reactions. The catalyst consisted of a mixture of copper(I)iodide, sodium ascorbate (SA), sulfonated bathophenanthroline (SBP) and 1,8-diaza-[5.4.0]bicycloundec-7-ane (DBU), all previously dissolved in a solvent mixture composed of H₂O/EtOH (1:1, v/v). This highly active “click” catalyst, when added to a mixture of the proper building blocks and reactant stoichiometry dissolved in a CH₂Cl₂/CH₃CN (7:3, v/v) organic solvent mixture, yielded rotaxane 18 and catenane 19 in respectable 75% and 57% yields, respectively. Most importantly, such an innovative synthetic approach allowed the preparation for the first time of interlocked D-A systems with the same molecular structural parameters. Therefore, the distinct molecular topology of catenanes and rotaxanes and its eventual effects in the photoinduced processes on the interlocked systems could be investigated [88,89,90,91,92,93,94,95,96].

A detailed electrochemical, steady-state and time-resolved absorption and emission investigation of rotaxane 18 and catenane 19, as well as of several model compounds, was carried out by Guldi’s group to determine all thermodynamic and kinetic parameters of the photoinduced processes (Figure 9). Not surprisingly, the sequence of events following exclusive photoexcitation of the ZnP moieties in rotaxane 18 and catenane 19 at 420 nm was identical to that observed for the ester-linked rotaxanes described in the previous section: (1) EnT on the ns time scale from ¹ZnP* to the ³MLCT* state of the [Cu(phen)₂]⁺; (2) ET from the ³MLCT* to C₆₀ to give the intermediate ZnP–[Cu(phen)₂]²⁺–C₆₀^•− CSS; (3) ET from the ZnP to the oxidized [Cu(phen)₂]²⁺ complex to give the final long-distance ZnP^•+–[Cu(phen)₂]⁺–C₆₀^•− CSS; (4) BET to regenerate the ground state [92,94]. This sequential ET processes were confirmed by electron paramagnetic resonance spectroscopy (EPR), in which the formation of both intermediate and final CSSs in rotaxane 18 (Figure 9a) and catenane 19 (Figure 9b) were detected [97,98].

However, the effects of the distinct molecular topologies of the rotaxane and catenane became clear in the investigation of the dynamics of the photoinduced processes. The lifetime of the final ZnP^•+–[Cu(phen)₂]⁺–C₆₀^•− CSS was 0.24 μs in rotaxane 18, while it was 1.10 μs in catenane 19 under the same conditions. From time-resolved fluorescence experiments, it was observed that the ¹ZnP* presented a biexponential decay with lifetimes of 61 ps and 400 ps for rotaxane 18, while catenane 19 showed a monoexponential decay with a lifetime of 500 ps for the ¹ZnP* excited state. Those findings informed that the ZnP stoppers in rotaxane 18 had distinctly electronic couplings with the other chromophores; therefore, they were at different distances from each other. A careful structural investigation by NMR spectroscopy revealed that rotaxane 18 was conformationally flexible and adopted a more compact structure in solution driven by secondary attractive π-π interactions between the chromophores. More specifically, one of the ZnP stoppers was closer to the [Cu(phen)₂]⁺ and C₆₀ subunits, while the other was further away. On the other hand, structural analysis of catenane 19 informed that it adopted an extended conformation with the chromophores as far apart as possible. Computation simulations confirmed the folded and extended conformations adopted by rotaxane 18 and catenane 19, respectively (Figure 10) [92,93,94].

To provide further evidence that the distinct lifetimes for the final ZnP^•+–[Cu(phen)₂]⁺–C₆₀^•− CSS were due to molecular topology, the authors investigated the photophysical properties of rotaxane 18 in the presence of 1,4-diazabicyclo [2.2.2]-octane (DABCO). The idea was to use formation of coordinative bonds between the ZnP stoppers and the bidentate DABCO ligand to disrupt the π-π interactions observed in 18 and, consequently, unfolding the rotaxane architecture to yield catenanelike structure 20 (Figure 11). Steady-state UV-Vis and fluorescence spectroscopic data revealed formation of the DABCO/ZnP complex with the expected 1:2 ratio, thereby confirming the expected transformation of rotaxane 18 into 20. Time-resolved transient absorption experiments on 20 indicated a monoexponential decay for the ZnP^•+ transient absorption at λ_max = 680 nm with a lifetime of 1.3 μs, which matched the long component decay observed for the biexponential rate law found for the fingerprint absorption of the C₆₀^•− species at λ_max = 1040 nm (the shorter component was attributed to BET in the ZnP–[Cu(phen)₂]²⁺–C₆₀^•− intermediate CSS). Accordingly, the final ZnP^•+–[Cu(phen)₂]⁺–C₆₀^•− CSS had virtually the same lifetime in 20 as in catenane 19. Based on those findings, the authors concluded that the longer lifetime observed for the final ZnP^•+–[Cu(phen)₂]⁺–C₆₀^•− CSS in catenane 19 was due to its more rigid structure, which kept the chromophoric subunits further apart, thus decreasing the electronic coupling of the radical ions when compared to that in the analogous but more flexible rotaxane 18. Thus, such extensive investigation clearly revealed, for the first time, the molecular topological effects on the dynamics of the photoinduced processes in interlocked molecules [94].

With those achievements, Schuster, Megiatto and Guldi focused their efforts to investigate the possibility of using the inherent circumrotation, shuttling, pirouetting and rocking motions of the entangled components to reposition the ZnP and C₆₀ subunits in order to favor ET, while retarding the wasteful BET to afford CSSs with longer lifetimes. As demonstrated by Sauvage’s works, one way to trigger intracomponent motions was removing the Cu(I) template ion from the central [Cu(phen)₂]⁺ complex [61]. However, the classical KCN treatment to remove the Cu(I) ion from Sauvage’s interlocked systems [5,19] was incompatible with the fullerene-based counterparts. From MALDI-TOF experiments on the fullerene-containing interlocked molecules, it was found that although the CN⁻ anions removed the Cu(I) ions from the rotaxanes and catenanes, unfortunately, they also reacted with the fullerene based compounds via nucleophilic attack that led to multiple additions of CN⁻ anions to the carbon cage. To circumvent this synthetic issue, the authors developed a demetallation protocol using NH₄OH as the active reagent, which proved to be successful in removing the Cu(I) ion template from the interlocked systems with no side reactions with C₆₀ [99].

Photophysical investigation of Cu-free rotaxane 21 (Figure 12a) yielded only ZnP and C₆₀ triplet excited states upon exclusively excitation of either chromophore. No evidence for ET reactions was found, even in high polar solvents. That was a surprising result as it was expected that removal of the Cu(I) ion from the rotaxane would allow a higher degree of freedom for the chromophores to reposition themselves close together, driven by π-π attractive interactions between the ZnP and C₆₀ groups. A structural investigation by NMR spectroscopy of rotaxane 21 was undertaken to identify the conformational changes brought by the removal of the Cu(I) template ion. As π-π interaction was thought to be the driving force of the conformational rearrangement, the shielding of specific protons in the rotaxane structure was used as a probe in the study. The NMR spectrum of 21 revealed that the protons of the phen-chelate on the ring component were shielded when compared to the analogous ones on the thread, thereby informing the researchers that the former were in the range of the ZnP magnetic anisotropy. Accordingly, the tweezerlike configuration of the ZnP stoppers in 21 favored formation of π-π attractive interactions with the planar phen moiety on the ring rather than with the spherical fullerene (Figure 12b), therefore preventing the expected pirouetting of the ring component about the thread. From the investigation, the authors demonstrated that removal of the Cu(I) template ion indeed promoted a conformational change in the rotaxanes; however, the resulting rearrangement of the chromophores did not improve the rates of the ET processes as originally thought, but rather shut them down completely [94]. On the other hand, the critical role of the [Cu(phen)₂]⁺ complex as mediator in the electronic interactions between the ZnP and C₆₀ groups in the interlocked molecules was confirmed [64,65,66,67,68,69,94].

5. Multi-Chromophoric Rotaxanes as Artificial Photosynthetic Models

Capitalizing on the synthetic achievements described in the previous sections, in conjunction with the elucidation of the sequential EnT and ET photoinduced processes in their ZnP–[Cu(phen)₂]⁺–C₆₀ interlocked molecules, Megiatto and Schuster decided to explore rotaxane architectures as molecular platforms to position several chromophores in space. The idea was to promote a cascade of directional and short range photoinduced EnT and ET processes among the interlocked chromophores to ultimately yield long-living CSSs as already observed by Moore, Moore and Gust in their covalently linked D-A systems [78,79,100,101,102,103,104,105,106]. By combining the [Cu(phen)₂]⁺-based stepwise template technique, ester reactions and “click” chemistry, the authors developed a synthetic tactic to prepare [Cu(phen)₂]⁺-based rotaxanes with distinct chromophores as stoppers (Figure 13). Accordingly, the strategy relied on a sequential “stoppering–threading–stoppering” approach. The hydroxyl group in phen-thread 23 was esterified with Zn(II)porphyrinate 22 using the EDC/DMAP coupling agent system to afford phen derivative 24. Threading of 24 through the C₆₀-based macrocycle by means of Sauvage’s Cu(I) template technique quantitatively yielded monostoppered pseudorotaxane 25, which was allowed to react with either alkynyl ferrocene or ethynyl Zn(II)phtalocyanine (ZnPc) under “click” conditions to produce the target multichromophoric rotaxanes 26 and 27 (Figure 13) [107].

A thorough investigation of the excited states’ properties and decay processes by steady state and time resolved emission spectroscopies, as well as transient absorption techniques, was carried out by Guldi’s group to reveal the photophysical properties of the new multichromophoric rotaxanes. Figure 14 shows the energy level diagrams that summarize the sequence and rate constants of the photophysical events occurring upon excitation of the mechanically linked chromophores. In the case of rotaxane 26 (Figure 14a), it was found that excitation of the ZnP group triggered the expected sequence of EnT and ET processes (steps 1, 3 and 4) to yield the Fc–ZnP^•+–[Cu(phen)₂]⁺–C₆₀^•− CSS with a lifetime of 2.3 μs, which was much longer than that observed for the same CSS in the parent (ZnP)₂–[Cu(phen)₂]⁺–C₆₀ rotaxane 18 (0.24 μs). That was a surprising result as it was expected on thermodynamic grounds that the Fc stopper would rapidly quench the ZnP^•+ via a charge shift reaction (step 6) to afford the final Fc^•+–ZnP–[Cu(phen)₂]⁺–C₆₀^•− CSS. Such a finding suggested that rotaxane 26 was also conformationally flexible. However, structural investigation by NMR spectroscopy to identify possible rotaxane conformations was unsuccessful, as unsymmetrical rotaxane 26 yielded interpretable NMR spectra in different solvents and temperatures. Fortunately, transient absorption spectra clearly revealed formation of the final Fc^•+–ZnP–[Cu(phen)₂]⁺–C₆₀^•− CSS in rotaxane 26, which mainly occurred from direct ET from the Fc stopper to the oxidized [Cu(phen)₂]²⁺ complex (step 5). The lifetime of the Fc^•+–ZnP–[Cu(phen)₂]⁺–C₆₀^•− CSS was 61 μs, which remains the longest lifetime ever detected in interlocked structures [107].

Regarding rotaxane 27, excitation of the ZnP stopper led to singlet-to-singlet EnT to the ZnPc moiety (step 1) in competition with intersystem crossing to produce ³ZnP* and ³ZnPc* (steps 2 and 3). The ¹ZnPc* manifold decayed via EnT to the [Cu(phen)₂]⁺ complex to produce the ³MLCT* (step 4), which in turn decayed via ET to the fullerene acceptor to yield the intermediate ZnP–ZnPc–[Cu(phen)₂]²⁺–C₆₀^•− CSS (step 5). A charge shift reaction from the ZnPc to the oxidized [Cu(phen)₂]²⁺ complex occurred (step 6) followed by a second charge shift reaction from the ZnP to the ZnPc^•+ radical cation (step 7) to yield the final ZnP^•+–ZnPc–[Cu(phen)₂]²⁺–C₆₀^•− CSS with a lifetime of 8.4 μs. The observed shorter lifetime of the final CSS in rotaxane 27 when compared to that of 26 was most likely due to the ZnP, [Cu(phen)₂]⁺ complex and ZnPc centered CSSs being energetically at a similar levels in rotaxane 27 (1.31. eV), while the Fc centered CSS was energetically favored (1.10 eV) in rotaxane 26. With those findings, Megiatto, Schuster and Guldi demonstrated that rotaxanes are useful molecular platforms to assemble several chromophores which are able to absorb light over a very wide range of the visible region, triggering a cascade of short-range EnT and ET processes to afford long-lived charge-separated states [107].

6. Active-Metal Template Synthesis of Photoactive Rotaxanes

There is a significant amount of literature on the transition metal template synthesis of interlocked molecules. Despite the richness of techniques available, complexes and geometries other than the tetrahedral [Cu(phen)₂] complex have not yet been fully explored in the synthesis of D-A mechanically linked systems [22,108]. However, few examples of D-A rotaxanes have been published recently, in which the interlocked systems are assembled via what has been called active metal template synthesis (AMT). The AMT technique has been popularized by Leigh and coworkers, who, in a series of elegant and influential studies, established the basis to explore the catalytic properties of transition metal ions rather than relying solely on the coordinative ones [22]. Accordingly, the metal ions not only organize the building blocks in an entwined arrangement but also actively promote the reaction that covalently captures the interlocked structure. The AMT tactic has proven to be a more powerful synthetic approach to assemble interlocked molecules [109,110,111,112,113,114,115,116,117,118,119] and impressive molecular machines, pumps and synthesizers [120,121,122,123,124].

In the field of mechanically linked artificial photosynthetic models, one elegant study by Weiss, Hyashi and Guldi groups has been recently published, in which the ZnP–C₆₀ rotaxane was assembled by the AMT technique (Figure 15) [125]. Specifically, treatment of a ZnP(II)porphyrinate-based macrocycle with [Cu(MeCN)₄][PF₆] led to formation of bimetallic complex 28, in which the appended bidentate phen moiety sequesters and positions the Cu(I) ion inside the cavity of the macrocycle. As Cu(I) ions are the promoters of “click reactions”, complex 28 is an endotopic catalyst that in the presence of alkynyl fullerene 29 and azide stopper 30 promotes the “click reaction” through the macrocycle cavity to yield target rotaxane 31 in 48% yield, after removal of the Cu(I) ions with EDTA. From structural investigations, the authors demonstrated that the resulting triazole ring on the thread was coordinated to the Zn(II) ion in the porphyrin core. This intracomponent interaction fixes the position of the fullerene stopper at about 5.6 Å to the ZnP moiety in rotaxane 31 (according to computational calculations) and prevents the shuttling movement of the ring along the thread.

The authors found that the photophysical decay of rotaxane 31 was dependent of the solvent polarity. In apolar toluene, excitation of the ZnP group yielded the ¹ZnP*, which decayed through ET to the C₆₀ to generate the ZnP^•+–C₆₀^•− CSS, and via intersystem crossing to produce the ³ZnP* manifold. The lifetime of the CSS in toluene was 12.8 ns, informing the authors that the ZnP group on the ring was anchored to the triazole moiety on the thread. Interestingly, in polar benzonitrile a short-lived (1.77 ns) and a long-lived (9.42 ns) CSSs were detected. The authors proposed that photooxidation of the ZnP by ET to the fullerene resulted in distortions of the ZnP scaffold, which in turn led to de-coordination of the triazole ring on the thread to allow benzonitrile solvent molecules to coordinate to the Zn(II) ion. Such a process allowed the ring to randomly move along the thread, thereby changing the distance between the chromophores to cause the observed biexponential decay of the ZnP^•+–C₆₀^•− CSS. Amusingly, it was found that the coordination of pyridine to ZnP group led to directional motion of the ring about the thread when the rotaxane was dissolved in apolar toluene, which brought the ZnP^•+ closer to the C₆₀^•−. Consequently, the lifetime of the ZnP^•+–C₆₀^•− CSS was biexponential in toluene, with lifetimes of 1.7 ns (major component) and 9.42 ns (minor component, corresponding to the fraction of rotaxanes in which the ZnP group was still anchored to the triazole moiety). Thus, this work demonstrated that formation of coordinative bonds can drive directional intercomponent motion in rotaxanes to influence the photoinduced processes in D-A interlocked systems.

Anderson’s group has also reported a series of complex photoactive interlocked structures [126,127,128,129]. In one of their designs [130], polyyne rotaxane 33 was assembled via the AMT technique (Figure 16). Phen-based macrocycle 32 is an endotopic catalyst for alkyne–alkyne coupling, which connects the stoppers together in an interlocked fashion to yield 33 in 65% yield, after removal of the Cu(I) ion. Due to the coordinating nature of the bidentate phen moiety, the authors also explored the rotaxane environment to position Re(I) complexes, which are highly luminescent, near the hexayne chain to form rotaxane 34. However, the luminescence of the Re(I) complex in 34 was totally quenched, which was a surprising finding as the HOMO-LUMO gap and the electron affinity of the polyyne chain were insufficient to thermodynamically allow EnT and ET processes between the components of rotaxane 34.

A thoroughly spectroscopic investigation associated with time-dependent Density Functional Calculations (TD-DFT) revealed the reasons for the unexpected luminescence quenching of the Re(I) complex in 34. Using ultrafast UV-NIR pump-probe transient absorption spectroscopy, it was found that a rapid triplet energy migration from the ³MLCT* excited state centered at the Re(I) complex to the thread hexayne occurred with a time constant of 1.5 ns. To understand the decay processes of the excited states on the polyyne thread, the authors investigated metal free rotaxane 33 by time-resolved infrared spectroscopy. Following the C≡C stretching frequencies (2100–2200 cm⁻¹) upon irradiation, it was found that the polyyne excited states (S_n, S₁ and T₁) had a significant cumulenic character, i.e., the C≡C triple bonds became longer and C–C bonds became shorter. The X-ray crystal structure of 34 revealed that the Re(CO)₃Cl center and the hexayne chain were in close contact, with the latter bending around the former. The authors concluded that such close contact caused the observed through-space electronic perturbation on both components in the excited state. This elegant work demonstrated the usefulness of rotaxane structures for investigating complex electronic excited interactions that would be rather difficult, if not impossible, to study with non-interlocked compounds [130].

7. Conclusions

This brief account concerning photoactive interlocked molecules assembled by transition metal template techniques describes major advances in the field over more than 30 years of research work. As one can see, mechanically linked D-A systems offer many advantages over D-A covalently linked analogs. Key parameters that are believed to be critical in natural photosynthesis, such as molecular topology and conformational changes of the photoactive subunits in the protein array upon photoexcitation, can be mimicked and better understood using photoactive interlocked molecules. Furthermore, the possibility of controlling the direction of the component molecular motions via external stimuli or stabilizing elusive intermediates renders those interactive photoactive systems very promising to further improve our understanding of ET and EnT processes. With the continuous advance of ultrafast spectroscopic techniques associated with the huge number of template synthetic tactics that can be used to prepare D-A interlocked molecules, we can safely conclude that this field has bright future.