Ruthenium(ii) Complexes Containing Functionalised Β-diketonate Ligands: Developing a Ferrocene Mimic for Biosensing Applications †

Three series of ruthenium complexes with the general formula Ru(bpy) n (β-diketonato) 3−n (bpy = 2,2'-bipyridine, n = 0, 1, 2) were prepared and investigated using cyclic voltammetry and UV-vis spectroscopy. Variation of both the number and electronic demand of the β-diketonato ligands resulted in well-defined modulation of the potential at which oxidation of the metal centre occurred. The observed potentials were shown to be in good agreement with calculated ligand electrochemical parameters. A novel ruthenium(II) complex with electrochemical behaviour similar to that of ferrocene was identified.


Introduction
][3][4][5][6][7][8][9] This is largely due to the fact that, in addition to its favourable electrochemical properties, the ferrocenyl group can be readily functionalised and is considered to be relatively stable in aqueous, aerobic media.However, the ferricenium ion formed following oxidation has been shown to decompose when exposed to chloride ions, [10][11][12][13] potentially limiting the application of ferrocene as a redox label in sensors designed for long-term analyte monitoring (e.g.implantable devices).
5][16][17][18][19][20][21][22] An advantage of utilising ruthenium species is that the potential at which oxidation of the metal centre occurs can be influenced by the electronic demand of the ligands occupying the primary coordination sphere.6][27][28][29][30] More subtle 'tuning' of the redox potential has previously been demonstrated by attaching either electron-withdrawing or electron-donating groups (EWG or EDG) to the peripheral sites around the coordinating ligands. 28,31,32re we present the synthesis and electrochemical characterisation of three series of ruthenium complexes (of the general formulae [Ru(bpy) 2 (β-diketonato)](PF 6 ), [Ru(β-diketonato) 3 ] and [Ru(bpy)(β-diketonato) 2 ], bpy = 2,2′-bipyridine) wherein the number and type of ligand is varied in order to tune the redox potential of the metal centre towards values suitable for biosensing applications.We take advantage of the fact that the potential at which oxidation of these ruthenium complexes occurs can be attenuated by varying the substituents on the β-diketonato ligands and identify several candidates for biosensor integration that have a very similar electrochemical profile to that of ferrocene.

Synthesis
Three series of ruthenium complexes with β-diketonato ligands were prepared (Scheme 1) where the complexes of Series I each contains a single β-diketonato ligand, Series II, three β-diketonato ligands and Series III, two β-diketonato ligands.
Synthesis of Series I [Ru(bpy) 2 (β-diketonato)](PF 6 ) complexes.Ruthenium complexes 1-7 (Scheme 1a) were prepared using a method adapted from a literature procedure. 33Compounds 1, 2 and 4 have been prepared previously (see ESI † for more information). 34All Series I complexes (excluding 4) were prepared by displacing chloride from Ru(bpy) 2 Cl 2 with selected β-diketones in the presence of a stoichiometric amount of t BuOK in a hot EtOH-H 2 O mixture.The desired product was precipitated following cooling and the addition of an aqueous solution of NH 4 PF 6 to the reaction mixture.Complex 4 was prepared by treatment of 1 with N-bromosuccinimide in CH 2 Cl 2 . 35ovel compounds 5, 6 and 7 were prepared in the same way as compounds 1-4 in 60%, 53% and 27% yields respectively.
Synthesis of Series II [Ru(β-diketonato) 3 ]complexes.9][40][41][42] Following the method of Collman et al., 42 nitration of 8 with Cu(NO 3 ) 2 in acetic anhydride was achieved to give Ru(NO 2 -acac) 3 (9) in 70% yield (Scheme 1b).Complex 8 also underwent facile bromination with N-bromosuccinimide to give Ru(Br-acac) 3 (10) in a 90% yield. 32Ru(I-acac) 3 (11) was synthesised in the same way using N-iodosuccinimide in 85% yield .Complex 12 was prepared in a 25% yield by following a literature procedure. 28,31ynthesis of Series III [Ru(bpy)(β-diketonato) 2 ]complexes.Due to the relatively substitution-inert nature of Ru(β-diketonato) 3 complexes, one approach to the synthesis of the mixed β-diketonato and bpy complexes is to prepare an intermediate that allows clean ligand substitution reactions with bpy. 43Ru-(β-diketonato) 2 (MeCN) 2 complexes were targeted as MeCN is sufficiently labile to be readily displaced by bpy.This method was originally applied to the synthesis of Ru(acac) 2 (bpy) (14). 44omplex 13 was synthesised in two steps via a bis-acetonitrile (Ru(dbm) 2 (MeCN) 2 ) (dbm = dibenzoylmethane) intermediate.This intermediate was accessed by the reduction of complex 12 with activated zinc dust in refluxing EtOH followed by cooling and the addition of excess MeCN.The mixture was then refluxed for a further 2 h resulting in a colour change from bright red to orange. 45Treatment of the intermediate (Ru(dbm) 2 (MeCN) 2 ) with an equimolar amount of bpy in refluxing EtOH gave Ru(dbm) 2 (bpy) (13) as a dark green solid in 60% yield.The methyl analogue (R 2 = Me) was prepared using the same procedure (8 was readily converted to 14 in a 60% overall yield).This method also proved successful for the preparation of nitro-analogue 15 but treatment of the tribromo/iodo ruthenium(III) complexes (10, 11) with elemental zinc resulted in dehalogenation of the ruthenium species and instead formed 14 along with other decomposition products.
Whilst this route initially appeared to be a more direct way of preparing Ru(bpy)(β-diketonato) 2 complexes compared with preforming the Ru(β-diketonato) 3 species first (Scheme 1b) our observation was that the overall yields tended to be significantly lower and in some cases the final product did not form at all.However this method did allow us to furnish sufficient amounts of complexes 16 and 17 for subsequent electrochemical analysis.

Cyclic voltammetry
All complexes in Series I-III were shown to undergo a redox cycle (assumed to be Ru 2+/3+ ) with a peak separation (ΔE) between 59 to 95 mV, and i pa /i pc at near unity indicating that the complexes were both electrochemically and chemically reversible.The E 1/2 of the complexes was independent of the scan rate (ν) and the redox processes were diffusion-controlled as i p vs. ν 1/2 was found to be linear.Waves were assigned by comparison to those of analogous metal complexes. 48ries I. Ruthenium(II) complexes 1-7 [Ru(bpy) 2 (β-diketonato)](PF 6 ).The potential at which oxidation of the ruthenium(II) complexes occurs in Series I (compounds 1 to 7) were observed to be subtly influenced by the various substituents on the β-diketonato ligands.Complex 2 has an electron-donating methyl group at the methine position (R 1 = Me, Scheme 1a).Consequently, a cathodic shift of the redox potential of 2 by 100 mV (128 mV vs. 228 mV) was observed when compared with that of complex 1 (Table 1).A similar cathodic shift was observed for the redox potential of 3 (R 1 = Et) although in this case the change is less pronounced (ΔE 1/2 = 47 mV) in line with the reduced electron donating capacity of the ethyl groups of 3 compared with the methyl groups of 2. These observations are in agreement with previous reports 49 and can be rationalised as follows: an electron-donating group (EDG) on the β-diketonato ligand increases the electron density around the metal centre and consequently destabilises electrons in the metal d-orbitals.Thus the metal centre is more readily oxidised resulting in the observed cathodic shift in the E 1/2 .
By the same argument, introducing an electron-withdrawing group (EWG) should result in an anodic shift in the E 1/2 and this effect was observed for complex 4 (R 1 = Br) when compared with 1 (ΔE 1/2 = 64 mV).
As expected, complex 5 (R 2 = i Pr) and complex 1 (R 2 = Me) had almost identical electrochemical profiles (ΔE 1/2 = 27 mV).The potentials at which complexes 6 and 7 were oxidised are quite similar (167 mV and 194 mV respectively) despite having functional groups with quite different electronic demands in the para position of the benzyl group (OMe vs. NO 2 ).This is not surprising considering that the aromatic ring is not in direct conjugation with the diketonate portion of the ligand bound to the ruthenium(II) ion.Consequently a modest cathodic shift (ΔE 1/2 = 61 mV for 6, ΔE 1/2 = 34 mV for 7) is observed in both cases due to the slight electron donating effect of the benzylic CH 2 group.
Series II and III.Ruthenium(II) complexes 8-17, Ru (β-diketonato) 3 vs.[Ru(bpy) (β-diketonato) 2 ](PF 6 ).The potentials at which all the Series II complexes (8-12) oxidised were all shifted cathodically compared with those of Series I due to the presence of three anionic ligands.In addition, as with Series I, the effect of a strongly EWG (R 1 = NO 2 , 9) results in a large anodic shift (671 mV) of the potential at which the complex oxidised when compared with 8.The halogenated analogues of Ru(acac) 3 (10 and 11) were also found to have an anodically shifted E 1/2 although the effect was less significant (ΔE 1/2 = 230 mV for Br, 231 mV for I).The potential at which complex 12 was oxidised was relatively close to that of 8 despite having six phenyl rings conjugated with the coordinating oxygen atoms (Table 2).
It might be expected that moving from Series II to Series III compounds, wherein one β-diketonato ligand is substituted for a bpy ligand, would result in a set of complexes with an E 1/2 closer to the desired redox window (−0.743V to +0.067 V vs. FcH +/0 , −0.343 V to +0.467 V vs. Ag/AgCl).
The replacement of a β-diketone in 12 and 8 by bpy to form 13 and 14 raised the E 1/2 by 490 mV and 672 mV respectively.This was attributed to the fact that bpy is a π-acceptorand the HOMO-LUMO gap of the complex could be increased and may result in a more stabilised Ru 2+ complex.The stabilising effect of bpy works in concert with the NO 2 -acac ligands used to form 15 resulting in an observed E 1/2 of −54 mV (vs.FcH +/0 ).In the same way it was possible to anodically shift the potential at which the ruthenium metal centre is oxidised by replacing the acac ligands with tri-and hexafluorinated 2,4-diones.A comparison of the E 1/2 s of 16 with 14 demonstrates this effect (ΔE 1/2 = 391 mV).The potential at which 17 oxidised was anodically shifted by 482 mV compared to that of 16.This difference corresponds to the number of trifluoromethyl substituents on the ligands: there are four in 17 compared to two in 16.Of the Series III complexes, 15 and 16 were each shown to have an E 1/2 very close to that of ferrocene (E 1/2 = −54 mV and −87 mV respectively).

Diffusion coefficient, D o
The diffusion coefficients, D o , of complexes 1-17 in acetonitrile were obtained using the Randles-Sevcik equation: where i p = current, n = number of electrons transferred, A = electrode area, D = diffusion coefficient, C = bulk concentration of redox species and ν = scan rate (Table 3).
The electrode area on the glassy carbon electrode, A, was determined using the known D o of ferrocene in acetonitrile with 0.1 M NBu 4 PF 6 as a supporting electrolyte (2.24 × 10 −5 cm 2 s −1 ). 50omplexes with bulky ligands, and those of higher charge, would be expected to have smaller D o values as they would be expected to migrate more slowly in solution. 15,51,52Additionally, changes in electron density around ruthenium brought on by donor/acceptor groups on β-diketonates will affect the degree to which counterions will attach to the reduced complexes. 53The structurally simplest complex in Series I, 1, has the largest D o while the rest of the complexes are between 1.182 to 1.561 × 10 −9 m 2 s −1 .However, the bulkiest complex of the series did not have the smallest D o .This could be attributed to the interaction between these charged complexes of Series I with acetonitrile and the supporting electrolytes.
In Series III, the D o for 13, which contains the bulkiest ligand, dbm, is about one order of magnitude smaller than that of 8.In the same way complex 17, with two trifluoromethyl groups on each β-diketonato ligand, has a smaller D o than 16 which only has one trifluoromethyl group on each ligand.Complex 15 has a lower D o when compared to 16 and 17, which can possibly be attributed to its interaction with the supporting electrolyte as the nitro group, although formally neutral, has considerable polarisation between its nitrogen and oxygen atoms.

Relationships between the nature of the ligands and E 1/2
The ruthenocycle formed by the coordination of acac-type ligands to Ru cations can be considered pseudo-aromatic.As such, the Hammett constant can be used to correlate the combined electronic effects of 'para' and 'meta' substituents on the β-diketonates with the potential at which the metal centres oxidised on the basis that such complexes can be thought of as having similar geometries and a common redox centre, for

Series I
][56] In this study correlation of the calculated Hammett constants with the recorded E 1/2 s is limited by the fact that the contribution of the bpy ligands is not included.Consequently, a reasonably linear correlation was observed for Series II and III but not for Series I (see ESI †).Instead a more comprehensive method for probing the relationship between the structural and electronic features of a complex and its redox potential was applied. 57The most commonly utilised additive model was originally developed by Lever 58 and has proved to be a useful tool for investigating metal-centred redox processes.The ligand electrochemical parameter, E L (L), used in this model and based on the Ru 2+/3+ redox couple, has been widely applied to examine the correlation between ligands and E 1/2 of metal complexes.For a metal (M) centre bound to multiple varying ligands (X x , Y y , Z z ), E L (L) is defined as: This formula can be used to predict the E 1/2 of complexes with an octahedral geometry. 58he ∑E (L) was calculated for all complexes in Series I-III where values for E (L) (L) were available (Table S4).A strongly linear correlation was established between E (L) (L) and E 1/2 of the complexes of Series I-III suggesting that the ligand contributions are additive (Fig. 1).

Correlation between UV-Vis absorbance and E 1/2
UV-vis spectra of complexes 1-7 (Series I) were recorded in MeCN at a concentration of 0.05 mM and the data is summarised in Table 4.In total, there are two strong absorption bands assigned to π→π* intraligand transitions in the UV region and three MLCT bands dπ (Ru 2+ )→π* (L) (L = bpy, β-diketonate) in the visible region.For complexes 1-3 the data reported here is in agreement with literature values. 48,59,60The UV-vis spectrum for complex 4 (R 1 = Br, R 2 = Me) shows a slight hypsochromic shift in the MLCT bands (504 vs. 515 nm, 558 vs. 568 nm) when compared with those of 1 that could be indicative of an increased HOMO-LUMO gap.This is in agreement with the electrochemical data as a notably higher potential is required to oxidise the Ru(II) metal centre of 4 (E 1/2 = 292 mV) when compared with 1 (E 1/2 = 228 mV).Overall, the absorbance spectra of all Series I complexes are dominated by the influence of the anionic β-diketonato ligand and are consequently very similar to each other (see Fig. S4 † for spectra). 61n contrast to Series I, the UV-vis absorption spectra of complexes in series III are markedly different from each other (Table 5).The absorptions due to the four phenyl rings attached to complex 13 are clearly visible in the UV-vis spectrum at 246 nm.These additional aromatic groups may also account for the broad band observed around 300 nm which could result from a degree of overlapping with the π→π* intraligand transitions of bpy.
The spectrum for complex 15 contains significant distortion of the band at around 300 nm where no clear maximum is visible.This could be attributed to an overlap of the nitro group absorption with the t 2g →π* (MLCT) band. 62The strongly electron-withdrawing nitro group may also account for the significantly blue-shifted MLCT maxima observed at 546 nm.
Both 16 and 17 contain strongly electron-withdrawing trifluoromethyl groups on the β-diketonato ligands.In line with this, the spectrum of complex 16 contains a MLCT band at 559 nm whereas the corresponding peak for 17 appears at 509 nm (Fig. 2 and 3).

Conclusions
In this paper, ruthenium complexes bearing bpy and β-diketonato ligands were prepared, fully characterised and investi-    ).The ligand electrochemical parameter was shown to have a linear relationship with E 1/2 for all three series.Of the complexes prepared, the E 1/2 of complexes 7, 13, 14, 15 and 16 were demonstrated to have E 1/2 s within the range considered suitable for biological sensors (−0.743V to +0.067 V vs. FcH +/0 ).Furthermore, the novel complex 15 and the previously reported complex 16 were demonstrated each to have an E 1/2 very close to that of ferrocene (E 1/2 = −54 mV and −87 mV vs. FcH +/0 , respectively) suggesting that they would be also good candidates for redox labels in electroactive biosensors.

Experimental methods
All manipulations of metal complexes and air sensitive reagents were performed using either standard Schlenk techniques or in a nitrogen/argon filled Braun glove-box.Reagents were purchased from Aldrich Chemical Company Inc. or Alfa Aesar Inc. and were used without further purification unless otherwise stated.For the purposes of air sensitive manipulations and in the preparation of air sensitive complexes, pentane, hexane, dichloromethane and tetrahydrofuran were dispensed from a PuraSolv solvent purification system.Solvents were dried and distilled under an atmosphere of nitrogen using standard procedures and stored under nitrogen in glass ampoules, each fitted with a Youngs © Teflon valve prior to use.Ethanol (EtOH) was distilled from diethoxymagnesium and dimethylformamide (DMF) was first dried over molecular sieves (4 Å) and distilled.Argon (>99.999%) was obtained from Air Liquide and used as received.Nitrogen gas for Schlenk line operation comes from in-house liquid nitrogen boil-off. 1H and 13 C NMR spectra were recorded on Bruker DPX300, DMX400, DMX500 and DMX600 spectrometers operating at 300, 400, 500 and 600 MHz ( 1 H) respectively and 75, 100, 125 and 150 MHz ( 13 C) respectively.Unless otherwise stated, spectra were recorded at 25 °C and chemical shifts (δ) are quoted in ppm.Coupling constants ( J) are quoted in Hz and have uncertainties of ±0.05 Hz for 1 H and ±0.5 Hz for 13 C. 1 H and 13 C NMR chemical shifts were referenced internally to residual solvent resonances.Deuterated solvents were purchased from Cambridge Stable Isotopes and used as received.Microanalyses were carried out at the Campbell Micro-analytical Laboratory, University of Otago, New Zealand or at the Research School of Chemistry, the Australian National University, Canberra, Australia.Mass spectra were acquired using a Thermo LTQ Orbitrap XL located in the Bio-analytical Mass Spectrometry Facility (BMSF) of the Mark Wainwright Analytical Centre, UNSW or on a Micromass ZQ (ESI-MS) mass spectrometer located in the School of Chemistry, UNSW.M is defined as the molecular weight of the compound of interest or cationic fragment for cationic metal complexes.
Cyclic voltammetry was performed using an Autolab PGSTAT 12 potentiostat (Eco Chemie, Netherlands).The CV data was processed using Nova Windows software.A conventional three-electrode electrochemical cell comprised of a working electrode (glassy carbon), a counter electrode ( platinum wire) and a reference electrode (aqueous or nonaqueous depending on the solvent system of the experiment) was used for all analysis.The aqueous reference electrode was a Ag/AgCl (3 M KCl) electrode (CH Instruments, Inc., TX, USA) while the non-aqueous reference electrode was a Ag/Ag + electrode (CH Instruments, Inc., TX, USA) which was referenced against ferrocene.The working electrodes used were glassy carbon electrodes (CH Instruments, Inc., TX, USA).The solutions for electrochemical measurements were deoxygenated with nitrogen gas for 10 min prior to measurements and kept under a blanket of nitrogen during the course of measurements.UV-Vis spectra were recorded on Shimadzu UV-2401PC in dry MeCN (5 × 10 −4 M) and reported as λ max /nm (ε/M −1 cm −1 ).
For characterisation of previously reported compounds see the ESI.† All complexes were isolated as a racemic mixture of Δ and Λ enantiomers.

Table 3
Diffusion coefficients for ruthenium complexes in Series I and III

Table 4
Absorption maxima, λ max , and molar extinction coefficients, ε for Series I