Synthesis and Light-Induced Actuation of Photo-Labile 2-Pyridyl-1,2,3-Triazole Ru(bis-bipyridyl) Appended Ferrocene Rotors

To realise useful control over molecular motion in the future an extensive toolbox of both actionable molecules and stimuli-responsive units must be developed. Previously, our laboratory has reported 1,1′-disubstituted ferrocene (Fc) rotor units which assume a contracted/π-stacked conformation until complexation of cationic metal ions causes rotation about the Ferrocene (Fc) molecular ‘ball-bearing’. Herein, we explore the potential of using the photochemical ejection of [Ru(2,2′-bipyridyl)2]2+ units as a stimulus for the rotational contraction of new ferrocene rotor units. Fc rotors with both ‘regular’ and ‘inverse’ 2-pyridyl-1,2,3-triazole binding pockets and their corresponding [Ru(2,2′-bipyridyl)2]2+ complexes were synthesised. The rotors and complexes were characterised using nuclear magnetic resonance (NMR) and ultraviolet (UV)-visible spectroscopies, Electro-Spray Ionisation Mass Spectrometry (ESI–MS), and electrochemistry. The 1,1′-disubstituted Fc ligands were shown to π-stack both in solution and solid state. Density Functional Theory (DFT) calculations (CAM-B3LYP/6-31G(d)) support the notion that complexation to [Ru(2,2′-bipyridyl)2]2+ caused a rotation from the syn- to the anti-conformation. Upon photo-irradiation with UV light (254 nm), photo-ejection of the [Ru(2,2′-bipyridyl)2(CH3CN)2]2+ units in acetonitrile was observed. The re-complexation of the [Ru(2,2′-bipyridyl)2]2+ units could be achieved using acetone as the reaction solvent. However, the process was exceedingly slowly. Additionally, the Fc ligands slowly decomposed when exposed to UV irradiation meaning that only one extension and contraction cycle could be completed.


Introduction
Exploring stimuli controlled motion in synthetic molecules will lead to advances in nanotechnology, and as such, molecular switches and machines have received increasing attention over recent years [1][2][3][4][5][6]. Relatively simple components, such as reversibly extendable/contractible units, using a stimulus as ubiquitous as light, could form useful components of larger assemblies. Many classes of molecular switches and machines have been realised, with molecular motion induced by a range of stimuli including chemical [7][8][9], electrochemical [10][11][12], and photochemical inputs [13][14][15]. Light-induced processes are particularly attractive as the input may be considered a clean energy source with excellent spatial resolution, and as such, photoactivated switching has been a focus of many research groups [16][17][18][19][20][21]. Some of the most notable examples of this subclass of molecular machines are those of Feringa and co-workers. They have developed a series of overcrowded alkenes model [Ru(pytri)(bipy) 2 ](BF 4 ) 2 complexes L1Ru and L2Ru were generated from the model ligands L1 and L2. The extended tetra-cationic rotor complexes L3Ru and L4Ru were then generated from the Fc rotor ligands L3 and L4. The four Ru(II) complexes were characterised using nuclear magnetic resonance (NMR), infrared (IR), and ultraviolet (UV)-visible absorption spectroscopies, along with electro-spray ionisation mass spectrometry (ESI-MS) and electrochemical investigations. Additionally, irradiation of the four Ru(II) complexes in acetonitrile solution with UV-light (254 nm) was shown to liberate [Ru(bipy) 2 (CD 3 CN) 2 ](BF 4 ) 2 , as shown through 1 H NMR spectroscopy and ESI-MS. Thermal re-coordination of the Ru(II) fragment(s) to complete the switching cycle was then attempted. However, the reverse process could not be completed because the Fc ligand decomposed during the photo-ejection process.

Synthesis of Ligands and Ru(II)(bipy) 2 Complexes
Ligands L1−L4 (Figure 1) were all synthesised using Pd(0)-catalysed Sonogashira cross coupling and Cu(I) catalysed azide-alkyne cyclo addition (CuACC) protocols. The syntheses of new ligands L2 and L4 are shown in the Supplementary Material (Schemes S1 and S2), along with full experimental procedures for all previously unreported intermediates. L1−L4 were characterised by 1 H NMR spectroscopy, 13 C NMR spectroscopy, ESI-mass spectrometry, elemental analysis, UV-visible spectroscopy, and in the cases of L2 and L4, X-ray crystallography (see Supplementary Material). Comparision of the 1 H NMR spectra of the model ligand with the rotor systems indicated that the Fc rotor based switches (L3 and L4) adopt a π-π stacked syn-conformation in solution. As shown in Figure 2, the 1 H NMR signals of the aromatic binding pockets of the disubstituted Fc ligands are shifted up-field relative to the corresponding signals in the model ligands, indicative of π-π interactions. The solid-state structures of L2 and L4 ( Figure 3) show in both cases the Fc units exhibit the expected eclipsed conformation of the cyclopentadienyl rings. The aromatic pytri binding pockets of L4 are also π-π stacked (pyridine-pyridine centroid distance = 3.58 Å, tri-tri centroid distance 3.913 Å and the angle between the ferrocene substituents = 6.58 • , (see Supplementary Material, Figure S53), providing further support for the contracted syn conformation of the disubstituted Fc ligands in their native state.  ). Furthermore, the two symmetric Fc cyclopentadienyl signals in the 1 H NMR of both L3 and L4 are replaced by a more complicated set of signals in the 1 H NMR of L3Ru and L4Ru, presumably due to the different isomeric mixtures expected with respect to the Ru(II) centres (∆/∆, ∆/Λ, and Λ,Λ). Diffusion-ordered 1 H NMR spectroscopy (DOSY NMR) was used to further confirm the numerous signals in the aromatic regions of the 1 H NMR spectra of L1Ru−L4Ru were, indeed, diffusing at the same rate, thus, belonging to species of the same molecular weight (Supplementary Material). Interestingly the aromatic signals in the 1 H NMR spectrum of L3Ru suggested the presence of two isomers (or two NMR equivalent pairs of isomers) in equal proportion, however, the same was not observed in the spectrum of L4Ru which displays a fully symmetric spectrum. Despite the potential for different isomeric mixtures (∆/∆, ∆/∆, and Λ,Λ), the DOSY and mass spectral data support only one molecular weight being present in each case (Supplementary Material).
Density Functional Theory (DFT) calculations (see the Supplementary Material for the details of the computational methods) were performed to support the notion the rotor complexes L3Ru and L4Ru must adopt an extended anti-conformation. The optimised open structure (dihedral angle ≈ ±180 • ) was observed to be the lowest energy conformation for both complexes ( Figure 5 and Supplementary Material, Figures S55−S62). As the dihedral angle is decreased, the relative energy increases (see Figure S54). In the ranges of ±180 • to ±110 • the relative energy is within 5 kJ mol −1 , and according to the Boltzmann distribution, will be accessible at room temperature. Dihedral angles of ±80 • (>7 kJ mol −1 ) or higher become energetically disfavoured, with populations of less than 6%, according to the Boltzmann distribution. This energy scan indicates that the ±180 • to ±110 • range will be most vastly populated, and smaller dihedral angles will be disfavoured. Thus, when the Fc rotors (L3 and L4) are uncomplexed, a syn-stacked conformation is preferred. However, complexation to the [Ru(bipy) 2 ] 2+ units causes a switch to an anti-extended conformation presumably the switching is driven by a combination of charge repulsion between the Ru(II) ion and destabilising steric interactions.

UV-Visible Spectroscopy
The UV-Visible spectra of all ligands (L1−L4) and complexes (L1Ru−L4Ru) were recorded in dichloromethane (DCM) solution ( Figure 6 and Supplementary Material, Figures S38 and S39). In the electronic absorption spectra of the ligands, the features of note are the π-π* transitions around 310 and 370 nm, and the transition at around 450 nm which is assigned as an Fc based process [68,69]. In the spectra of the complexes, convoluted absorbance features between 400 and 470 nm are observed (along with shoulder features at around 500 nm) and are assigned as Ru based Metal-to-Ligand Charge Transfer (MLCT) transitions. For L1Ru and L2Ru, the extinction coefficient(s) for these transitions are~10 and~13 × 10 3 L mol −1 cm −1 (at 422 and 428 nm) respectively. Likewise, for L3Ru and L4Ru, extinction coefficients of~24 and~26 × 10 3 L mol −1 cm −1 (at 422 and 427 nm) are observed respectively. The ratio of Ru(II) ions to Fc units (1:1 for L1Ru and L2Ru, and 2:1 for L3Ru and L4Ru) correlates with the approximate two-fold increase in the intensity of the transition(s) assigned to ruthenium-based MLCT transitions. The UV-Visible absorption spectra of L1Ru−L4Ru were also recorded in acetonitrile (see Supplementary Material, Figure S40) and differ little to the spectra recorded in DCM.   Table 1 below. The current and potential observed for all processes were reproducible over multiple cycles. L1−4 displayed the expected Fc + /Fc couple between 0.62 and 0.75 V. The two ferrocenyl oxidation potentials for the mono-substituted ligands are slightly lower than those displayed by the di-substituted ligands ( Table 1). All four of L1Ru−L4Ru displayed the Fc + /Fc couple along with a Ru III/II couple and two reduction processes assigned to the two bipy ligands on each Ru(II) centre. As displayed by the ligands, the ferrocenyl based oxidation process occurs at a lower potential in the mono-nuclear complexes versus the di-nuclear analogues. As expected, the current response for the Ru III/II couple is larger for L3Ru and L4Ru versus L1Ru and L2Ru, consistent with the molecular formulae. The ruthenium based oxidation process occurs at a higher potential in the 'inverse' pytri complexes (L2Ru and L4Ru) than in the 'regular' pytri analogues, similar to previously reported related compounds [49]. Furthermore, both bipy reduction processes are also shifted to the anodic potential in the 'inverse' analogues, relative to the 'regular'. Reduction processes for the pytri ligands were not observed under the conditions employed here within the solvent window.    (Figure 7). The explanation as to why L4Ru reaches a fully ejected state in less than half the time taken for the same process in L2Ru is not entirely clear, but we suggest two plausible reasons: L4Ru was able to absorb more photons in the appropriate energy range of the light source used, per Ru(II) centre, or perhaps once one Ru(II) centre is lost from L4Ru, the free pytri binding pocket may have become involved in the second ejection process.

Electrochemistry
The behaviour of the reg-pytri analogues was significantly different to that of the inv-analogues. After 26 days of irradiation in [D3]acetonitrile solution, L1Ru had undergone approximately 75% conversion to the photoreaction products (based on the integration of key 1 H NMR signals). Signals that did not correspond to the starting Ru complex, nor [Ru(bipy) 2 (CD 3 CN) 2 ] 2+ or the free ligand began to grow at the beginning of the irradiation experiment. These signals were presumed to be intermediate ligand ejection products (Scheme 1 below), where either the 1,2,3-triazole (intermediate a) or the pyridine (intermediate b) of the ferrocene ligand had been replaced by an acetonitrile solvent molecule, but not the other. This is consistent with the observation of at least two sets of these unassigned signals having similar chemical shifts and multiplicity appearing and is also consistent with previously reported photo ejection experiments of pytri ligands [49]. Interestingly, after the same length of time (26 days), L3Ru had reached a point where essentially none of the original signals of the intact L3Ru species were present. However, full conversion to [Ru(bipy) 2 (CD 3 CN) 2 ] 2+ had also not occurred, as a significant amount of the material appeared to be in an intermediate stage. When analogous [Ru(bipy) 2 (reg-Bnpytri)] 2+ complexes (where reg-Bnpytri = 2-(1-benyzl-1H-1,2,3-triazol-4-yl)pyridine), which were not substituted with Fc, were irradiated (254 nm, [D3]acetonitrile), no photo-reaction was observed [49]. We propose a different absorption profile due to the substituents of the pytri ligands presented here as the likely cause of this difference.
With the photochemical behaviour of the four ruthenium complexes under UV irradiation determined, we turned our attention to the reverse process of thermal re-coordination. L2Ru had appeared in the respective mixtures. The proportion of L1Ru and L2Ru grew overtime with continued heating. After 12 days the 1 H NMR of the mixture of L1 and [Ru(bipy) 2 (CH 3 CN) 2 ](BF 4 ) 2 was almost identical to that of L1Ru. The equivalent process was observed for L2, however at a lower rate. This showed that the thermal re-coordination of at least L1 or L2 to [Ru(bipy) 2 (CH 3 CN) 2 ] 2+ was, in fact, possible, but not from the mixture produced by UV-irradiation in [D3]acetonitrile, due to the decomposition of the ligands post Ru(II) photo-ejection.
The equivalent re-coordination experiments in [D6]acetone for L3 and L4 were not attempted, as the ejection from L3 was extremely slow, and both have a very low solubility in acetone, making neither an ideal candidate for reversibly switchable rotors.
While the current Fc switches are unstable under the photo-switching conditions, the results obtained herein suggest that ruthenium(II) based ligand photo-election reactions could be exploited to develop reversible photo-switches in correctly designed, more robust systems. Efforts towards new more rapid and robust ruthenium(II) based photo-switches based on other "click" 1,2,3-triazoles ligands are underway.

Materials and Methods
Unless otherwise stated, all reagents were purchased from commercial sources and used without further purification. 1,1 -diiodoferrocene was synthesised and purified according to a literature preparation [70]. The ligands L1 and L3 were synthesised using our previously reported methods [67]. Solvents were laboratory reagent grade. Substances  (0.1 M, 100 mL). The organic layer was separated and the aqueous phase extracted with DCM (4 × 50 mL). The combined organic layers were washed with brine (80 mL), dried over Na 2 SO 4 , filtered, and the solvent was removed under reduced pressure. The residue was purified by column chromatography (silica gel, gradient DCM, then 9:1 DCM/acetone) to give an orange solid upon removal of solvent. The product was further purified by precipitation from DCM with petrol, collection by filtration, and desiccating. Yield  General procedure for synthesis of [Ru(bipy) 2 ] 2+ complexes of L1-L4: [Ru(bipy) 2 (Cl) 2 ] (1.2 eq. for L1 and L2, and 2.2 eq. for L3 and L4) and AgBF 4 (2.0 eq./Ru ion) were combined in degassed acetone (typically 5 mL) in the dark, and stirred at room temperature for two hours. The resulting AgCl precipitate was filtered off through celite, and the solution was added to a degassed acetone solution (typically 10 mL) of the desired ferrocene ligand under argon. The mixture was then refluxed in the dark overnight. The resulting solution was filtered through Celite and the solvent removed under reduced pressure. The residue was taken up in the minimum amount of DCM, precipitated by addition of diethyl ether, filtered over a vacuum and rinsed with diethyl ether (2 × 5 mL). The air-dried products were then further dried in a desiccator for at least 24 h.  All cyclic voltammetry (CV) experiments were performed in solutions at 20 • C (DCM) with a concentration of 1 mM of electroactive analyte and 0.1 M NBu 4 PF 6 as the supporting electrolyte. A three-electrode cell was used with Cypress Systems 1.4 mm diameter glassy carbon working, Ag/AgCl reference and platinum wire auxiliary electrodes. Voltammograms were recorded with the aid of a Powerlab/4sp computer-controlled potentiostat (AD instruments, Castle Hill, NSW, Australia). Potentials for all complexes were referenced to the reversible formal potential (taken as E • = 0.00 V) of the [Fc*] +/0 redox couple of decamethylferrocene [71]. In the cases of L1-L4, scans between −2.0 V and 1.7 V revealed no processes other than the ferrocenyl oxidation.
All photochemical experiments were carried out in a custom built UV photochemical reactor (at the University of Otago, Dunedin, NZ, USA), containing four UV-C lamps (Rayonet RPR-2537A, Southern New England Ultraviolet Co., Branford, CT, USA) emitting monochromatic 254 nm radiation, arranged symmetrically around the perimeter of the photoreactor. Samples were suspended in the centre of the UV-bulb arrangement, and an in-built fan was used to maintain a constant temperature of 35 • C during irradiation experiments.

Supplementary Materials:
The following are available online. Full experimental data of all the synthetic intermediates in the synthesis of L2 and L4. Additionally, 1 H and 13 C NMR, ESI-MS, and UV-visible spectral data, CV and DPV plots, and X-ray crystallographic data for the compounds are provided. The computational methods are also detailed.

Funding:
The authors wish to thank the University of Otago for funding. J.D.C. thanks the MacDiarmid Institute for Advanced Materials and Nanotechnology for additional financial support.