Cobalt Complexes with Redox-active Anthraquinone-type Ligands

Three anthraquinone-type multidentate ligands, HL 1-3 (HL = 2-R-1 H -anthra[1,2-d]imidazole-6,11-dione, HL 1 ; R = (2-pyridyl), HL 2 ; R = (4,6-dimethyl-2-pyridyl), HL 3 ; R = (6-methoxy-2-pyridyl)), were prepared, and their complexation behaviour were investigated. Three bis-chelate cobalt complexes with the formula [Co II (L 1-3 ) 2 ] . n(solv.) ( 1 , 2 , and 3 for HL 1 , HL 2 , and HL 3 , respectively), in which the ligands adopted tridentate binding modes, were synthesized and structurally characterized by single-crystal X-ray analyses. Electrochemical studies of 1 - 3 in CH 2 Cl 2 reveal three reversible redox waves, assigned to ligand and cobalt-centred processes. Further complexes were obtained in which HL 1 adopted a bidentate binding mode, stabilising the mono-chelate [Co II (HL 1 )(NO 3 ) 2 (DMF) 2 ] ( 4 ) species and tris-chelate [Co III (L 1 ) 3 ] ( 5 ) complex in which the cobalt ion was in its 3+ state. The electrochemical properties of complex 5 were investigated in DMF, and the Co(II)/Co(III) redox couple was found to have negatively shifted compared to that of complex 1 , while the ligand-based processes became irreversible. Tridentate chelation is found to stabilise the anthraquinone ligands and unlocks their redox multi-stability.


Introduction
Redox-active ligands, and in particular, non-innocent ligands allow the development of functional coordination compounds with varied and tuneable electronic properties. 1 Indeed, transition metal complexes with redox-active ligands can be successfully employed as reaction sites for a wide range of catalytic molecular conversions. 2 In addition, compounds consisting of redox active metal ions and ligands with similar redox potentials can show significant switchable physical properties arising from intramolecular electron transfer induced by external stimuli (temperature, pressure, light, etc.). 3 From the catalysis perspective, Tanaka et al. reported the reduction of carbon dioxide using ruthenium-quinone systems, 4 while Chang et al. discovered proton-coupled hydrogen evolution systems based on an o-phenylene diamine iron compound. 5 In the development of switchable systems, valence tautomerism was observed in cobalt-dioxolene complexes, and a range of more complicated systems have been reported, 6 some of which were developed by following a rational molecular design approach. 7 Metal dithiolene compounds have been the most studied system due to the strength of the electronic delocalisation between the transition metal centres and the non-innocent ligands. 8 These systems exhibit electrical conductivity based on the formation of a hybrid conduction band comprising ligand p orbitals and metal d orbitals, and have been studied intensively due to their potential applications in molecular electronics.
A substantial body of research in recent years has focussed on advanced materials based on organic donor-acceptor pairs, with systems based on the donor tetrathiafulvalene (TTF) and acceptor tetracyanoquinodimethane (TCNQ) among the most famous and useful for the construction of functional molecular materials. 9 Their analogues have been employed to modify the redox properties of the system and allowed crystal engineering toward functional molecular devices. 10 Electron transfer complexes have attracted attention from the perspective of organic molecular conductors, organic light emission devices and solar batteries, and in order to develop functional molecular compounds, it is important to explore a range of redox-active ligands and to study their complexation behaviour toward metal ions, the potential catalytic applications of such complexes, and the dramatic changes to their electronic states triggered by external stimuli. The molecular design and exploration of redoxactive ligands and their complexes underpins this area of research, and such studies will lead to a clear pathway to multifunctional molecular materials.
We previously investigated functional polynuclear complexes based on redox-active ligands incorporating triphenyl amine and ferrocene groups, and their magnetic and electrochemical properties were studied. 11   pyrazole), showed intramolecular antiferromagnetic interactions and ligand-centered redox processes.
Both trimethylamine and ferrocene groups can be oxidized to yield stable cationic moieties, and can be used to significantly alter the properties of the molecular assemblies. The combination of multiple redox-active moieties in the same molecular system is a key target towards the generation of functional molecular materials. With this in mind, we focused on the combination of redox-active anthraquinone-type ligands and cobalt ions. The anthraquinone moiety shows two reversible reduction steps at -0.70 and -1.46 V (versus SCE). 12 Examples of coordinative anthraquinone ligands and their complexes have shown that the redox properties of the organic moiety are preserved in the molecular assemblies. 13 In this work, three multidentate anthraquinone-based ligands were synthesised (Scheme 1) and employed to obtain five cobalt complexes. The structural, electronic, electrochemical and luminescent properties of the ligands and their complexes were studied.

Materials and general methods
All the reagents were commercially purchased and used without further purification.

Syntheses of ligands
The anthraquinone-type ligand HL 1 (Scheme 1) was prepared according to literature, 14 and HL 2 and HL 3 were synthesized by a similar method to that used for HL 1 as described below.
Data collections were performed on a Bruker SMART APEX II for all complexes, with a CCD area detector with graphite monochromated Mo-K ( = 0.71073 Å) radiation. All structures were solved by direct methods and refined by fullmatrix least-squares methods based on F 2 using the SHELXL software. 15 Non-hydrogen atoms were refined anisotropically. All hydrogen atoms were positioned geometrically and refined with isotropic displacement parameters according to the riding model. All geometrical calculations were performed using the SHELXL software.

Physical measurements
Infrared (IR) spectra were recorded (400-4000 cm -1 region) on a SHIMADZU FT-IR 8400 spectrometer using KBr pellets. Cyclic voltammetry measurements were carried out in a standard one-compartment cell under N2 at 20 o C equipped with a platinum-wire counter electrode, an SCE reference electrode, and a glassy carbon (GC) working electrode using a BAS 620A electrochemical analyser. The measurements were performed in CH2Cl2 or DMF (for complex 5) with 0.1 M tetra-nbutylammonium hexafluorophosphate (n-Bu4NPF6) as the supporting electrolyte. Analyte concentrations for all ligands and complexes were 0.6 mM. In the measurement of HL 1 in DMF, a concentration of 1.0 mM was used. The E values were references to the E1/2 value of ferrocene, which was measured under identical conditions each time. UV-vis absorption spectra were recorded on SHIMADZU UV-3150 spectrometer. Fluorescence spectra were measured by JASCO FP-8500 spectrofluorometer at room temperature. Absolute quantum yields were determined using an integrating sphere photometer. Measurements were performed using degassed solvent.
When using HL 1 , the mono-dentate [Co II (HL 1 )(NO3)2(DMF)2] (4) and tris-chelate cobalt complexes [Co III (L 1 )3] (5) were obtained by slight modification of the synthetic method. The ligands have two coordination modes, the bidentate N2 mode of pyridine benzimidazole chelate, and the tridentate N2O mode in which the additional quinone oxygen is also binding. The benzimidazole moiety of the ligands has a dissociative proton, thus the ligands can be neutral or monoanionic. This flexibility allows the ligands to stabilise complexes in a range of different ways.
In order to elucidate the electrochemical character of HL 1 , a zinc complex, [Zn II (L 1 )2] (6), was synthesized as a reference. Complexes 1, 4 and 5 were bis-, mono-, and tris-chelate complexes with the same ligand HL 1 , respectively. Selective syntheses of these complexes were achieved by optimization of the reagent ratio and selection of solvent. The mono-chelate complex 4 was synthesized without base in DMF solution, which is a coordinative solvent. Complexes 1 and 5 can be obtained by the same synthetic method, but can be separated by their different solubilities.

Molecular structures
The molecular structures of complexes 1-6 were determined by single-crystal X-ray diffraction analysis. Complexes 1-3, which have a bis-chelate mode, have similar structures. As an example, the structural features of complex 1 are described here in detail. 1 crystalized in triclinic space group P1 and consists of two HL 1 ligands, one cobalt ion and solvent molecules in the lattice. The ligands coordinate the cobalt ion in the tridentate mer manner. The cobalt ion has an octahedral coordination geometry of N4O2 donor atoms, of which two are pyridine and two are imidazole nitrogen atoms and the two oxygen atoms are donated by the anthraquinone moiety. The average coordination bond length is 2.163(10) Å. Therefore, the cobalt ion is divalent based on BVS calculations 17 1-3, it is considered that the ligands are deprotonated based on charge balance. X-ray structural analysis was also performed for the control complex 6. The molecular structure of 6 is almost identical to complex 1. Single crystal X-ray analysis reveals that 4 consists of one ligand HL 1 , one cobalt ion, two nitrate ions and one DMF molecule. The ligand coordinates the cobalt ion in a bidentate fashion. The cobalt ion has octahedral coordinate geometry with two nitrogen atoms from HL 1 and four oxygen donor atoms from the nitrate and DMF groups. Considering the coordination bond lengths, BVS calculations 14 , and charge balance, the cobalt ion in 4 can be assumed to be divalent and the ligand HL 1 protonated.   (18) 10.334 (4) 10.0343 (17) 13.525 (6) 12.720 (3) 10.681(3) b (Å) 17.818 (3) 11.306 (4) 11.1530 (18) 13.060 (6) 15.676 (3) 10.758(2) c (Å) 16.549 (2) 14.456 (6) 17.350 (3) 17.173 (8)  Complex 5 is a tris-chelate system, in which the bidentate site of the ligand coordinates the cobalt ion, and the ligands are deprotonated and monoanionic. The C=O bond length of the ligand is slightly short (1.216(4) -1.222(4) Å), compared to that of protonated ligand. There are no counter anions, and BVS calculations confirm that the cobalt ion is in its trivalent state. 14 The average coordination bond lengths, 1.938(3) Å, are shorter than those of complexes 1-4. Each ligand has two kinds of nitrogen donor atoms; pyridyl and imidazolyl. The ligands are orientated around the cobalt ion such that the three pyridine donors coordinate to cobalt ions in the mer mode, as do the benzimidazole moieties.

Electrochemical properties
Electrochemical properties of the three ligands and complexes 1-3, 5 and 6 were investigated. The data collected for 6 was used as a control sample. Cyclic voltammetry of the ligands and complexes was performed in CH2Cl2 solution and recorded relative to a SCE reference electrode.
The resultant voltammograms are shown in Fig. 4, and the redox potentials (E1/2) and assignments are provided in Table 1. Figure 4 Cyclic voltammograms of ligand HL 1-3 (dotted lines, red, blue and green for HL 1 , HL 2 and HL 3 , respectively), and 1-3 (solid lines, red, blue and green for 1, 2 and 3, respectively). Open circuit potentials are indicated by the black crosses, and scan directions by the arrows. According to the literature, anthraquinone shows two reversible reduction process at -1.13 and -1.89 V versus Fc/Fc + in NMP, which are attributed to Quinone (Q) / Semiquinone (SQ • ) and Semiquinone (SQ • ) / Catecolate (Cat 2 ) redox waves, respectively. 18 HL 1 shows two reduction waves at -1.29 and -1.49 V. The first reduction process is slightly more positive than that of anthraquinone due to the substituent effect of the benzimidazole moieties. HL 2 and HL 3 show similar twostep reduction processes. Differences between the redox properties of the different ligands are negligible, indicating that the influence of the pyridine substituents on the quinone moiety is weak. Complexes 1, 2, and 3 show three quasi-reversible redox waves, two of which can be assigned to ligand-based processes, and the other to the reversible oxidation of the cobalt centre. The redox processes around -1.18 ~ -1.36 V can be assigned as Quinone (Q) / Semiquinone (SQ • ), while the waves at -1.66 ~ -1.57 V can be considered as Semiquinone (SQ • ) / Catecholate (Cat) processes. The redox wave of the ligand-based Q/SQ • process in the voltammograms of 1 and 2 is slightly split due to intra-molecular electronic communication between the two ligands through the cobalt centre, suggesting that the molecule is electronically conjugated through the metal ion. The SQ • /Cat 2 processes were negatively shifted compared with the free ligands. The reason for this negative shift is likely due to the dianionc charge of the two-electron reduced complexes. The oxidation potentials of the cobalt ion appeared in the range of 0.38 -0.70 V, with the oxidation peak of 1 falling at more negative potential than that of 2 and 3. Considering the electron donation effects of the substituent groups, which would typically result in a negative potential shift in the redox chemistry of a coordinated metal centre due to the increased ligand basicity, the tendency of the potential shift is the opposite of that expected. The positive shifts observed in the cobalt redox in compounds 2 and 3 can be explained by examining the structural features of the complexes. Note that coordination bond lengths about cobalt(II) ions tend to be shorted as the oxidation to cobalt(III) ions. There is substantial steric hindrance introduced by the substituent groups in 2 and 3, and this will disfavour the required structural changes, resulting in a positive shift in the oxidation potential of the cobalt ion. In the case of complex 5, irreversible redox waves were observed in the potential range of -1.02 ~ -1.73 V, which are assigned to Q/SQ • and SQ • /Cat 2 processes, accompanied with protonation or decomposition. A Co(II)/Co(III) redox process was observed at -0.06 V, which is close to the redox potentials of previously reported cobalt complexes with tris azole heterocycle ligands. 19 Cyclic voltammetry of complex 6, the zinc derivative with HL 1 , was measured for comparison (Supporting info, Figure S1). There were three reduction waves attributed to ligand reduction processes, at similar potentials to those of complex 1. 6 | J. Name., 2012, 00, 1-3 This journal is © The Royal Society of Chemistry 2012

Luminescent properties
9,10-anthraquinone shows luminescence at 412 nm, and HL 1 shows emission at 562 nm. 16 The three ligands, HL 1 , HL 2 and HL 3 , show similar absorption and broad fluorescence to the reported one (Figure 6a). Broad transitions around 350-450 nm for the ligands (395, 398 and 405 nm for HL 1 , HL 2 and HL 3 , respectively) can be assigned as ILCT (= intra-ligand charge transfer) transitions, while the emission bands at 519, 523 and 540 nm for HL 1 , HL 2 and HL 3 , respectively, are derived from ILCT excited states. 17 The absolute quantum yields of HL 1 , HL 2 and HL 3 were determined to be 0.95, 1.75 and 4.24 %, respectively. The similarity of the general spectroscopic features of the three ligand derivatives indicates that substituent groups on the pyridine ring exert minimal influence on the electronic properties of the anthraquinone moiety. The luminescence properties of complexes 1, 5 and 6 were investigated and these spectra are shown in Figure 6b. Complex 1 shows relatively weak emission around 500 nm, while complexes 5 and 6 show stronger emission around 550-600 nm. The reason for the weak emission of 1 can be understood as paramagnetic quenching due to the Co(II) centre and effect of dissociation equilibrium of the complex.  Soc. Rev., 2013, 42, 1440-1459ACS Catal., 2012, 2, 270-279. shifts of the emission bands of 5 and 6 were caused by the decrease in LUMO energy effected by the charge and acidity of the coordinated metal ion. Luminescence spectra for 1-3 are shown in Figure S2, and exhibit small differences depending on the spectroscopic properties of HL 1 -HL 3 respectively. These tuneable luminescent properties and solution stability may lead to potential applications in future switchable photochemical systems.