Diastereoisomers As Probes for Solvent Reorganizational Effects On

Chemical Physics 324 (2006) 8–25 www.elsevier.com/locate/chemphys

Diastereoisomers as probes for solvent reorganizational eﬀects on IVCT in dinuclear ruthenium complexes

Deanna M. DAlessandro, F. Richard Keene *

Department of Chemistry, School of Pharmacy and Molecular Sciences, James Cook University, Townsville, Qld. 4811, Australia

Received 19 June 2005; accepted 1 September 2005 Available online 3 October 2005

Abstract

5+ 0 0 IVCT solvatochromism studies on the meso and rac diastereoisomers of [{Ru(bpy)2}2(l-bpm)] (bpy = 2,2 -bipyridine; bpm = 2,2 - bipyrimidine) in a homologous series of nitrile solvents revealed that stereochemically directed specific solvent effects in the first solvation shell dominated the outer sphere contribution to the reorganizational energy for intramolecular electron transfer. Further, solvent proportion experiments in acetonitrile/propionitrile mixtures indicated that the magnitude and direction of the specific effect was dependent on the relative abilities of discrete solvent molecules to penetrate the clefts between the planes of the terminal polypyridyl ligands. In particular, the specific effects were dependent on the dimensionality of the clefts, and the number, size, orientation and location of the solvent dipoles within the interior and exterior clefts. 5+ 5+ IVCT solvatochromism studies on the diastereoisomeric forms of [{Ru(bpy)2}2(l-dbneil)] and [{Ru(pp)2}2(l-bpm)] {dbneil = dibenzoeilatin; pp = substituted derivatives of 2,20-bipyridine and 1,10-phenanthroline} revealed that the subtle and systematic changes in the nature of the clefts by the variation of the bridging ligand, and the judicious positioning of substituents on the terminal ligands, profoundly influenced the magnitude of the reorganizational energy contribution to the electron transfer barrier. 2005 Elsevier B.V. All rights reserved.

Keywords: Solvatochromism; Intervalence charge transfer; Dinuclear; Stereochemistry; Ruthenium; Solvent reorganization

1. Introduction The particular appeal of mixed-valence complexes of the II III 5+ form [{LnM }(l-BL){M Ln}] (M = metal centers, Dinuclear ligand-bridged mixed-valence complexes have L = terminal ligands and BL = bridging ligand) is the played a pivotal role in the assessment of activation barri- observation of an absorption band in the near-infrared ers to intramolecular electron transfer since the disclosure (NIR) region of the electronic spectrum which is identified of the Creutz–Taube ion, [{Ru(NH3)5}(l-pyz){Ru- as the optically induced intervalence charge transfer 5+ (NH3)5}] (pyz = pyrazine), in 1973 [1]. Systems of this (IVCT) transition. IVCT measurements provide a sensitive genre have provided important experimental insights into and powerful probe to elucidate aspects of intramolecular the roles of solvent dynamics [2–15], ion-pairing [16–21], electron transfer processes as the energy (mmax), intensity encapsulation [22,23], temperature [24–28], and redox (e) and bandwidth (Dm1/2) of these transitions can be quan- asymmetry [29,30], and they have been used as model sys- titatively related to the factors which influence the activa- tems to verify the salient predictions of several important tion barrier to electron transfer [31,32]. theoretical models that describe the activation barrier to For symmetrical, valence-localized mixed-valence sys- electron transfer [31–35]. tems, Hush [31,32] proposed the relationship 0 mmax ¼ ki þ ko þ DE ; ð1Þ 0 * Corresponding author. Tel.: +61 0 7 47814433; fax: +61 0 7 47816078. where DE represents the energy contributions due to E-mail address: [email protected] (F.R. Keene). spin–orbit coupling and/or ligand field asymmetry, and

0301-0104/$ - see front matter 2005 Elsevier B.V. All rights reserved. doi:10.1016/j.chemphys.2005.09.016 D.M. DAlessandro, F.R. Keene / Chemical Physics 324 (2006) 8–25 9 ki and ko are the inner- and outer-sphere reorganizational The elucidation of the relative contributions of contin- (Franck–Condon) parameters, respectively: ki corresponds uum and non-continuum effects is the subject of consider- to the energy required for reorganization of the metal–li- able experimental interest in the attempt to develop more gand and intra-ligand bond lengths and angles, and ko is sophisticated theoretical models for solvent reorganiza- the energy required for reorganization of the surrounding tional contributions to the electron transfer barrier [2,40]. solvent medium. The solvent contribution is generally Dielectric continuum theory obscures the ‘‘molecularity’’ modelled as a one-dimensional classical mode due to the of the solvent by neglecting the properties of individual sol- low frequencies of the coupled vibrations [31–34]. The vent molecules, and this underpins the recent theoretical spherical cavity dielectric continuum model given by Eq. interest directed towards understanding the molecular basis (2) provides a framework for the calculation of the sol- of reorganizational effects [2,40]. vent reorganizational contribution in which the electron The experimental strategy for extracting information at donor and acceptor are modelled as two non-interpene- the molecular level using IVCT solvatochromism studies trating spheres, embedded in the dielectric continuum involves probing the first solvation shell separately from [31,32]. the bulk solution. Dinuclear ruthenium complexes incorpo- 1 1 1 1 rating ammine and cyano ligands have been extensively k ¼ e2 . ð2Þ o a d D D investigated in this regard because of the existence of op s strong directional H-bonding and donor–acceptor interac- The parameters a and d define the molecular radii and dis- tions between the chromophore ligands and individual sol- tance between the donor and acceptor, e is the unit elec- vent molecules [6,11–15,41]. These specific solvent tronic charge, and Ds and Dop are the macroscopic static interactions coexist with, and often dominate dielectric and optical dielectric constants of the solvent, respectively. continuum effects [2]. Correlations have been found be- In accordance with Eqs. (1) and (2), mmax should vary line- tween the IVCT solvent shifts and empirical solvent param- arly with the solvent dielectric function (1/Dop 1/Ds), eters such as the Gutmann donor and acceptor numbers 2 0 with slope e (1/a 1/d) and intercept ki + DE at (1/ [42]. In studies of dinuclear ruthenium mixed-valence com- Dop 1/Ds) = 0, and when the length of the bridging li- plexes based on –Ru(NH3)5,–trans-Ru(NH3)4(py) and 0 gand is varied (at fixed a) mmax should vary linearly with –Ru(bpy)(NH3)3 with pyz, 4-cyanopyridine and 4,4 -bpy 2 1/d, with slope e (1/Dop 1/Ds) in a given solvent. bridging ligands [19,43,44], the IVCT energies correlate lin- While the predictions of the dielectric continuum model early with the Gutmann solvent donor number (DN) due have been consistent with the results from a number of to specific H-bonding interactions between the ammine li- IVCT solvatochromism studies of mixed-valence dinuclear gands and the solvent molecules. In each case, the magni- ruthenium complexes [2,36,37], the model breaks down tude of the specific interaction increased with the donor when the underlying assumptions of the classical model number of the solvent, and the number of NH3 ligands in are invalidated, or in the presence of specific solvent–solute the chromophore. interactions or dielectric saturation effects [2]. Theoreti- IVCT solvatochromism studies in solvent mixtures have cally, the dielectric continuum model is formulated in terms demonstrated that the solvent reorganizational process oc- of a one-dimensional classical mode for the solvent, but it curs predominantly within the first solvation layer, and is inadequate for systems which exhibit coupled high-fre- may be profoundly influenced by the systematic replace- quency quantum modes which must be explicitly treated ment of individual solvent molecules in the immediate through a quantum mechanical approach [15]. Eq. (2) also vicinity of the mixed-valence chromophore neglects the volume occupied by the donor and acceptor [2,5,7,13,14,19,45]. (the ‘‘excluded volume’’) and is valid only when the dis- There is clearly a need for experimental studies of IVCT tance between the redox centers exceeds the sum of their ra- solvatochromism which provide insights into the micro- dii (d 2a). The corrections due to non-spherical fields scopic solvent reorganizational contributions to the intra- around the metal centers become increasingly important molecular electron transfer barrier. as the distance between the metal centers is decreased. Experimentally, the analysis of IVCT solvatochromism 1.1. Scope and objectives of the present study data according to the spherical (and ellipsoidal [13,15,38,39]) dielectric continuum models has often been The majority of experimental IVCT studies have been severely confounded by non-continuum effects. These is- conducted by varying global features of the complexes, sues have been addressed in an extensive review of medium such as the identities of the bridging and terminal ligands, effects on the IVCT properties of mixed-valence complexes and the constituent metal centers. In addition, the theoret- by Chen and Meyer [2], and include specific solvent–solute ical implications of the results have often been complicated interactions and dielectric saturation effects, in addition to by ion-pairing, and ambiguities in the geometries of the ion-pairing contributions from the counter and electrolyte complexes due to a lack of structural rigidity and/or stereo- ions, the concentration of the chromophore and the chem- isomeric purity [46,47]. ical oxidant used for the generation of the mixed-valence In the present study, the investigation of the IVCT sol- complex. vatochromism in the mixed-valence diastereoisomeric 10 D.M. DAlessandro, F.R. Keene / Chemical Physics 324 (2006) 8–25

5+ 0 forms of [{Ru(pp)2}2(l-bpm)] (pp = 2,2 -bipyridine and uents on the terminal polypyridyl ligands – while maintain- its derivatives; bpm = 2,20-bipyrimidine) provides a new ing the identity and coordination environments of the experimental approach to probe the microscopic origins component metal centers. For symmetrical complexes, ki of solvent reorganizational effects on intramolecular elec- and DE0 are expected to be identical for the diasterereoi- tron transfer processes. As an example, the dinuclear spe- somers of the same complex, so IVCT solvatochromism 4+ 0 cies [{Ru(bpy)2}2(l-bpm)] (bpy = 2,2 -bipyridine) exists studies of the diastereoisomeric forms permit a direct probe in two diastereoisomeric forms – meso, KD (DK) and of stereochemically induced ko variations on the intramo- racemic (rac) (illustrated in Fig. 1), the latter comprising lecular electron transfer barrier. This may provide new the enantiomeric pair DD and KK. While the identity and and intimate insights into solvent reorganizational effects coordination environments of the metal centers are identi- at the molecular level. cal in each diastereoisomeric form, a significant difference We report details of a three-pronged approach to this may be discerned in the nature of the ‘‘clefts’’ formed be- problem: tween the planes of the terminal bpy ligands [47]. ‘‘Interior clefts’’ are formed between the bpy ligands immediately (a) IVCT solvatochromism studies. An investigation was ‘‘above’’ and ‘‘below’’ the plane of the bridging ligand: undertaken on the IVCT solvatochromism for the 5+ the terminal ligands are approximately orthogonal in the diastereoisomeric forms of [{Ru(bpy)2}2(l-bpm)] meso diastereoisomer and parallel in the rac form. In addi- in the homologous series of the nitrile solvents aceto- tion, ‘‘exterior clefts’’ are evident between the planes of the nitrile {CH3CN; AN}, propionitrile {CH3CH2CN; terminal bpy ligands at either end of the complex and are PN}, n-butyronitrile {CH3(CH2)2CN; BN}, iso- i identical for two diastereoisomeric forms. butyronitrile {(CH3)2HCCN; BN} and benzonitrile The variation in the dimensions of the clefts between the {C6H5CN; BzN}. The macroscopic properties of the diastereoisomeric forms of the same complex may have sig- solvents (as defined by the solvent parameter 1/ nificant consequences for differential solvent and anion Dop 1/Ds) vary over the range 0.5127 (AN) to association at the molecular level. Indeed, the differential 0.3897 (BzN), while the subtle variation in the molec- association of eluent anions such as toluene-4-sulfonate ular shape, size and symmetry through the series of fCH3ðC6H4ÞSO3 g gives rise to the separation of the diaste- aliphatic and aromatic nitriles allows for a detailed reoisomeric forms in the chromatographic cation-exchange analysis of the microscopic origins of the solvent reor- separation process [46–49]. ganizational energy due to stereochemically directed The diastereoisomers of symmetrical dinuclear polypyr- solvent effects. idyl complexes exhibit several attractive features over dinu- (b) Influence of the bridge on the selectivity of solvent asso- clear complexes which have been employed to date for the ciation. The IVCT solvatochromism properties of 5+ investigation of reorganizational contributions to the [{Ru(bpy)2}2(l-dbneil)] (dbneil = dibenzoeilatin, IVCT processes: the complexes are structurally rigid, and Fig. 2) were examined to assess the effect of increasing the dimensions of the clefts may be varied in a subtle and the dimensions of the interior clefts by comparison systematic way through stereochemical variation, bridging with the analogous diastereoisomers of [{Ru(b- 5+ ligand modification, or the judicious positioning of substit- py)2}2(l-bpm)] . (c) Influence of the terminal ligands on the selectivity of solvent association. This study involved the systematic modification of the dimensions of the interior and exterior clefts by the judicious positioning of substituents on the terminal ligands in the series 5+ 0 [{Ru(pp)2}2(l-bpm)] , where pp = 5,5 -dimethyl- 0 0 0 0 2,2 -bipyridine (5,5 -Me2bpy), 4,4 ,5,5 -tetramethyl- 0 2,2 -bipyridine (Me4bpy), 2,9-dimethyl-1,10-phenan- 0 0 throline (Me2phen) and 4,4 -di-tert-butyl-2,2 -bipyri- t dine ( Bu2bpy), shown in Fig. 3. In each case, the IVCT characteristics of the separated diastereoisomers were investigated in the homologous series of nitrile solvents.

2. Experimental

2.1. Materials Fig. 1. (a) Front view and (b) top view of the meso (KD) and rac (DD) 4+ diastereoisomers of [{Ru(bpy)2}2(l-bpm)] illustrating the subtle variation in the dimensions of the clefts above and below the plane of the Hydrated ruthenium trichloride (RuCl3 Æ 3H2O; Strem, bridging ligand. Hydrogen atoms are omitted for clarity. 99%), 2,20-bipyrimidine (bpm; Lancaster), 2,20-bipyridine D.M. DAlessandro, F.R. Keene / Chemical Physics 324 (2006) 8–25 11

Fig. 2. The bpm and dbneil bridging ligands with the crystallographically determined Ru–Ru distances in their dinuclear complexes (meso- 4+ 4+ [{Ru(Me2bpy)2}2(l-bpm)] [48] and meso-[{Ru(bpy)2}2(l-dbneil)] [68]).

Fig. 3. Terminal polypyridyl ligands used in the present study.

0 0 (bpy; Aldrich, 99+%), 4,4 -dimethyl-2,2 -bipyridine (Me2b- Acetonitrile (AN or CH3CN; Aldrich, 99.9+%), and 0 0 0 py; Aldrich), 5,5 -dimethyl-2,2 -bipyridine (5,5 -Me2bpy; propionitrile (PN; Aldrich) were distilled over CaH2 before Aldrich), 2,9-dimethyl-1,10-phenanthroline (Me2phen; use, while acetone (BDH, HPLC grade) was distilled over 0 0 t Monsanto), 4,4 -di-tert-butyl-2,2 -bipyridine ( Bu2bpy; K2CO3 and dichloromethane over CaCl2 prior to use. n- i Aldrich, 98%), stannous chloride (SnCl2 Æ 2H2O; Ajax), Butyronitrile (BN; Aldrich, 99+%), iso-butyronitrile ( BN; lithium chloride (LiCl; Aldrich, 99+%), ammonium hexa- Aldrich) and benzonitrile (BzN; Aldrich) were used as re- fluorophosphate (NH4PF6; Aldrich, 99.99%), potassium ceived. N,N-Dimethylformamide (DMF; Ajax) was dis- hexafluorophosphate (KPF6; Aldrich, 98%), lithium tetrakis- tilled under reduced pressure [50] (76 C at 39 mmHg) (pentafluorophenyl)borate diethyletherate (Li{B(C6F5)4} Æ immediately prior to use. 0 0 0 Et2O; Boulder Scientific), ethylene glycol (Ajax, 95%), 4,4 ,5,5 -Tetramethyl-2,2 -bipyridine (Me4bpy) was sup- sodium benzoate (Aldrich, 98%), sodium toluene-4-sulfo- plied by Dr. Bradley Patterson (JCU) and was prepared nate (Natos; Aldrich, 98%), DOWEX 1 · 8, 50–100 mesh according to the literature method [51,52]. (Aldrich) and Amberlite IRA-400 (Aldrich) anion exchange resins (Cl form) and laboratory reagent solvents 2.2. Instrumentation and physical methods were used as received. Tetra-n-butylammonium hexaflu- 1 orophosphate ([(n-C4H9)4N]PF6; Fluka, 99.9+%) was dried 1D and 2D H NMR spectra were performed on a Var- in vacuo at 60 C prior to use and ferrocene (Fc; BDH) was ian Mercury 300 MHz spectrometer. The 1H NMR chem- purified by sublimation prior to use. SP Sephadex C-25 ical shifts for all complexes are reported relative to 99.9% (Amersham Pharmacia Biotech), and silica gel (200–400 d3-acetonitrile {CD3CN; Cambridge Isotope Laboratories mesh, 60 A˚ , Aldrich) were employed for the chromato- (CIL)} at d = 1.93 ppm. 1H NMR assignments were per- graphic separation and purification of ruthenium com- formed with the assistance of COSY experiments to iden- plexes [47]. tify each pyridine ring system. 12 D.M. DAlessandro, F.R. Keene / Chemical Physics 324 (2006) 8–25

Elemental microanalyses were performed at the Micro- K1=2 P ¼ c ð3Þ analytical Unit in the Research School of Chemistry, Aus- 2 þ K1=2 tralian National University. For selected complexes, an c allowance for hydration was necessary to account for anal- Spectral deconvolution of the NIR transitions was per- ysis figures within the acceptable limits (±0.4%). formed using the curve-fitting subroutine implemented Electrochemical measurements were performed under within the GRAMS 32 commercial software package. argon using a Bioanalytical Systems BAS 100A Electro- For the dinuclear complexes, convergence of the iteration chemical Analyzer. Cyclic and differential pulse volta- procedure was generally achieved for three gaussian- mmograms were recorded under Ar in 0.02 M shaped bands under the IVCT manifold. Based on the reproducibility of the parameters obtained from the decon- [(n-C4H9)4N]{B(C6F5)4}/CH3CN at +25 C using a glassy carbon working electrode, a platinum wire auxiliary elec- volutions, the uncertainties in the energies (mmax), intensi- trode and an Ag/AgCl (0.02 M [(n-C H ) N]{B(C F ) }/ ties {(e/m)max} and bandwidths (Dm1/2) were estimated as 4 9 4 6 5 4 1 1 1 CH CN) reference electrode. Ferrocene was added as an ±10 cm , ±0.0001 M and ±10 cm , respectively. In 3 2 internal standard on completion of each experiment {the all cases, the correlation coefficient (R ) for the fits reported ferrocene/ferrocenium couple (Fc+/Fc0) occurred at was >0.995. +550 mV vs. Ag/AgCl}: all potentials quoted in millivolts vs. Fc+/Fc0 [53]. Cyclic voltammetry was performed with 2.4. Synthetic procedures a sweep rate of 100 mV s1; differential pulse voltammetry was conducted with a sweep rate of 4 mV s1 and a pulse Microwave-assisted syntheses were conducted in a amplitude, width and period of 50 mV, 60 ms and 1 s, round-bottom flask fitted with condenser, mounted within respectively. Potentials from DPV experiments are a modified microwave oven (Sharp, Model R-2V55; 600W, reported ±3 mV. In order to obtain reasonable electro- 2450 MHz) on medium-high power [60–62]. A detailed ac- chemical responses, measurements in the 0.02 M count of the column cation-exchange chromatographic procedures for separation of the diastereoisomers of dinu- [(n-C4H9)4N]{B(C6F5)4}/CH3CN electrolyte required a concentration of complex which was approximately clear species has been reported previously [48,49]. [(n-C H ) N]{B(C F ) } was obtained by metathesis double that in 0.1 M [(n-C4H9)4N]PF6/CH3CN. iR com- 4 9 4 6 5 4 pensation was not employed for the electrochemical from tetra-n-butylammonium tetrakis(pentafluorophe- measurements. nyl)borate ([(n-C4H9)4N]{B(C6F5)4}) using an adaption of the literature procedure [63], which is described in Supple- 2.3. UV/Vis/NIR spectroelectrochemistry mentary Material. The mononuclear ruthenium complexes [Ru(DM- Electronic spectra were recorded using a CARY 5E SO)4Cl2] [64], cis-[Ru(pp)2Cl2] Æ 2H2O {pp = bpy, Me2bpy} UV–visible–NIR spectrophotometer interfaced to Varian [65] and cis-[Ru(Me4bpy)2Cl2] [66] were obtained according 0 WinUV software. The absorption spectra of the electrogen- to the literature procedures. [Ru(5,5 -Me2bpy)2Cl2] was erated mixed-valence species were obtained in situ by the synthesized using an adaption of the ‘‘ruthenium blue’’ use of an optically semi-transparent thin-layer electrosyn- method reported by Togano et al. [65], and [Ru(Me2- t thetic (OSTLE) cell (path length 0.685 mm) mounted in phen)2Cl2] and [Ru( Bu2bpy)2Cl2] were prepared via a the path of the spectrophotometer [54,55]. modification of a procedure reported by Sullivan et al. Solutions for the solvent proportion experiments were [67]: the details of syntheses and characterizations of these prepared by serial dilution of solvent mixtures containing three [Ru(pp)2Cl2] species are provided in the Supplemen- acetonitrile and propionitrile. To minimize artefacts in tary Material. the NIR spectral data due to ion-pairing and concentration effects which are known to influence the IVCT transitions 2.4.1. [Ru(Me2phen)2(bpm)](PF6)2 of dinuclear complexes [20,21,28,56,57], the spectra were [Ru(Me2phen)2Cl2] (500 mg, 0.850 mmol) and bpm measured using a constant concentration of complex (123 mg, 0.776 mmol) were heated at reflux in 3:1 EtOH/ 3 3 (0.40 · 10 M) in 0.02 M [(n-C4H9)4N]{B(C6F5)4}at water (200 cm ) for 5.5 h during which time the solution +25 C. Spectroelectrochemical experiments were repeated attained an orange coloration. Upon cooling, the mixture three times at each concentration, and the results reported was diluted with distilled water (50 cm3) and loaded onto as an average of the triplicate experiments. a column (15 cm · 3.5 cm) containing SP Sephadex C-25 The ‘‘raw’’ absorption spectra (e vs. m) were scaled as support. Separation of the mononuclear product from e(m)/mdm [31,58] and all data reported in both tabulated the crude mixture was achieved via a gradient elution pro- and graphical format are presented as e/m vs. m. cedure using aqueous 0.1–0.4 M NaCl as the eluent. The Due to comproportionation of the mixed-valence spe- orange band of mononuclear material was precipitated cies [59] a correction for the concentration of the species as the PF6 salt from the eluate by addition of a saturated present in solution was determined from Eq. 3. In all cases, solution of aqueous KPF6. The solid was isolated by vac- the proportion, P of the complex in the mixed-valence form uum filtration and washed with diethyl ether. Yield: (at equilibrium) was >97.5%. 433 mg (53%). Anal. Calculated for C36H30F12N8P2Ru: D.M. DAlessandro, F.R. Keene / Chemical Physics 324 (2006) 8–25 13

C, 44.8; H, 3.11; N, 11.6%; Found: C, 44.8; H, 3.05; N, sealed and the substrate recycled several times down its 11.5%. length with the aid of a peristaltic pump. The ‘‘effective column length’’ (ECL) – which represents the length of t 2.4.2. [Ru( Bu2bpy)2(bpm)](PF6)2 support travelled by the sample for visual band separation A suspension of bpm (120 mg, 0.760 mmol) in ethylene – was 180 cm. The two dark green bands were collected 3 glycol (3 cm ) was heated in a modified microwave oven and precipitated as the PF6 salts by addition of a saturated on medium-high power for 20 s to complete dissolution. solution of KPF6. The solid products from each band were t [Ru( Bu2bpy)2Cl2] (700 mg, 0.988 mmol) was added and purified on silica gel: the complex was dissolved in a mini- the mixture heated at reflux for a further 5 min during mum volume of acetone and loaded onto a short column of which time the solution attained an orange coloration. silica gel (200–400 mesh equilibrated with AR acetone; ca. Upon cooling, the mixture was diluted with distilled water 3 cm in length · 1.5 cm in diameter), washed alternately (50 cm3) and loaded onto a column (15 cm · 3.5 cm) con- with copious amounts of water and acetone, and then taining SP Sephadex C-25 support. Separation of the eluted with AR acetone containing 5% NH4PF6. Following mononuclear product from the crude mixture was achieved the addition of an equal volume of water and removal of via a gradient elution procedure using aqueous 0.1–0.4 M the acetone under reduced pressure, the precipitate was iso- NaCl as the eluent. The orange band of mononuclear mate- lated by vacuum filtration, washed with chilled water and 1 rial was precipitated as the PF6 salt by addition to the elu- diethyl ether and dried in vacuo. The H NMR character- ate of a saturated solution of aqueous KPF6. The solid was istics of the diastereoisomers are provided in the Supple- isolated by vacuum filtration and washed with diethyl mentary Material. ether. Yield: 609 mg (74%). Anal. Calculated for 0 C44H54F12N8P2Ru: C, 48.7; H, 5.01; N, 10.3%; Found: C, 2.4.4. [{Ru(5,5 -Me2bpy)2}2(l-bpm)](PF6)4 0 48.6; H, 4.99; N, 10.3%. [Ru(5,5 -Me2bpy)2Cl2] (200 mg, 0.370 mmol) and bpm [{Ru(Me2bpy)2}2(l-bpm)](PF6)4 was synthesized as (26.6 mg, 0.168 mmol) were heated at reflux in ethylene gly- 3 described previously [48,49], and [{Ru(bpy)2}2(l- col (2 cm ) in a modified microwave oven according to the dbneil)](PF6)4 was supplied by Prof. Moshe Kol [68]. De- method previously reported for [{Ru(Me2bpy)2}2(l- tails of the separation of these complexes into their diaste- bpm)](PF6)4 [48,49] on medium-high power for 10 min. reoisomeric forms and the 1H NMR characterization of the The separation of the orange mononuclear material from diastereoisomers have been reported previously [48,49,69]. the desired dark green product was achieved via a gradient elution procedure as described previously for [{Ru(b- 2.4.3. [{Ru(bpy)2}2(l-bpm)](PF6)4 py)2}2(l-bpm)](PF6)4. Yield: 226 mg (85%). Anal. Calcu- [Ru(bpy)2Cl2] Æ 2H2O (300 mg, 0.5765 mmol) and bpm lated for C56H56F24N12P4Ru2: C, 40.1; H, 3.37; N, 10.0%; (41.4 mg, 0.262 mmol) were heated at reflux (120 C) in Found: C, 40.0; H, 3.34; N, 10.0%. 10% water/ethylene glycol (20 cm3) for 5 h. Upon cooling, The separation, isolation and purification of the diaste- the dark green solution was diluted with distilled water reoisomeric forms was achieved as described above, but (50 cm3) and loaded onto a column (dimensions: using 0.25 M sodium benzoate solution as the eluent 15 cm · 3.5 cm) containing SP Sephadex C-25 support. (ECL 50 cm). Separation of the dinuclear product from the crude mixture [{Ru(Me4bpy)2}2(l-bpm)](PF6)4 was synthesized, and was achieved via a gradient elution procedure using aque- the diastereoisomers separated and isolated, in an analo- 0 ous 0.1–0.5 M NaCl as the eluent. An orange band of gous manner to that described above for [{Ru(5,5 -Me2b- mononuclear material eluted first (0.2–0.3 M NaCl), fol- py)2}2(l-bpm)](PF6)4. lowed by a dark green band of the dinuclear material (0.5 M NaCl), which was precipitated as the PF6 salt by 2.4.5. [{Ru(Me2phen)2}2(l-bpm)](PF6)4 addition of a saturated aqueous solution of KPF6, isolated [Ru(Me2phen)2bpm](PF6)2 (150 mg, 0.155 mmol) and by vacuum filtration, washed with chilled water and diethyl [Ru(Me2phen)2Cl2] (160 mg, 0.250 mmol) were heated in ether, and dried in vacuo at 40 C for 4 h. Yield: 385 mg ethylene glycol (5 cm3) in a modified microwave oven on (94%). The 1H NMR spectrum of the dinuclear product medium-high power for 10 min. Upon cooling, the resul- was identical to that reported previously [70]. tant dark green solution was diluted with distilled water The diastereoisomeric mixture was converted to the (50 cm3) and loaded onto a column (15 cm · 3.5 cm) con- chloride salt by stirring an aqueous suspension with DOW- taining SP Sephadex C-25 support. Separation of the dinu- EX 1 · 8 anion exchange resin (50–100 mesh; Cl form). clear product from the crude mixture was achieved via a The complex was sorbed onto a chromatography column gradient elution procedure using aqueous 0.1–0.5 M NaCl (ca. 1 m in length · 1.6 cm in diameter) containing SP as the eluent. An orange band of mononuclear material Sephadex C-25 cation exchanger as the support, and sepa- eluted first, followed by the desired dark green product, ration of the diastereoisomers was achieved using aqueous which was precipitated as the PF6 salt by addition of a sat- 0.25 M sodium toluene-4-sulfonate solution as the eluent urated solution of aqueous KPF6. The solid was isolated by [48]. The separation of the stereoisomers was not achieved vacuum filtration and washed with diethyl ether. Yield: in a single passage down the column, so the column was 208 mg (76%). Anal. Calculated for C64H54F24N12P4Ru2: 14 D.M. DAlessandro, F.R. Keene / Chemical Physics 324 (2006) 8–25

C, 43.4; H, 3.05; N, 9.48%; Found: C, 43.3; H, 3.01; N, reaction with an excess of [Ru(pp)2Cl2]. This probably re- 9.40%. flects the unfavorable steric interactions between the t The separation, isolation and purification of the diaste- methyl substituents in the Bu2bpy and Me2phen terminal reoisomeric forms was achieved as described above, but ligands in the formation of the dinuclear complex. using 0.20 M sodium benzoate solution as the eluent. Separation of the diastereoisomeric forms of [{Ru- t 4+ [{Ru( Bu2bpy)2}2(l-bpm)](PF6)4 was synthesized, and (pp)2}2(l-bpm)] was achieved by cation-exchange chro- the diastereoisomers separated and isolated, in an analo- matography using SP Sephadex C-25 support with aqueous gous manner to that described above for [{Ru(Me2- sodium toluene-4-sulfonate or sodium benzoate as the elu- phen)2}2(l-bpm)](PF6)4. ent [46–49]. The latter was found to provide a more effi- The detailed description of these syntheses and separa- cient separation (i.e., shorter ECL) for the complexes 0 tion of the diastereoisomeric forms of all the new dinuclear incorporating alkylated terminal ligands (5,5 -Me2bpy, t species is provided in the Supplementary Material, together Me4bpy, Bu2bpy and Me2phen). with the details of the 1H NMR characterizations of the The final products from the synthetic procedures were diastereoisomers. purified chromatographically and investigated by NMR spectroscopy and electrochemistry for confirmation of their 3. Results and discussion structural identity and purity [48,49].

3.1. Diastereoisomer synthesis, separation and structural 3.2. 1H NMR studies characterization Structural characterizations of the meso and rac forms The complexes [{Ru(bpy) } (l-bpm)]4+ and [{Ru- 2 2 of [{Ru(pp) } (l-bpm)]4+ {pp = bpy, 5,50-Me bpy, Me b- (Me bpy) } (l-bpm)]4+ have been synthesized previously 2 2 2 4 2 2 2 py, Me phen, tBu bpy} were performed using one- and [30,48,49,70–77], either by a thermal method involving 2 2 two-dimensional (1H COSY) NMR techniques, and are de- the reaction of the cis-[Ru(bpy) Cl ] Æ 2H O precursor with 2 2 2 cribed in the Supplementary Material. bpm in ethylene glycol under reflux, or a microwave- assisted synthetic procedure. In the present case, the latter 4+ method was used for the complexes [{Ru(pp)2}2(l-bpm)] 3.3. Electrochemistry and electronic spectroscopy 0 {pp = bpy, Me2bpy, 5,5 -Me2bpy, Me4bpy}, which re- sulted in comparable yields to those obtained from the The electrochemical properties of the diastereoisomeric 4+ 0 thermal methods but with a significant reduction in the forms of [{Ru(pp)2}2(l-bpm)] {pp = bpy, 5,5 -Me2bpy, t reaction times (typically 10 min rather than 2 h for the ther- Me4bpy, Me2phen and Bu2bpy} and [{Ru(bpy)2}2- mal procedure). (l-dbneil)]4+ were investigated by cyclic and differential 4+ t For the complexes [{Ru(pp)2}2(l-bpm)] {pp = Bu2b- pulse voltammetry in acetonitrile containing 0.1 M py and Me2phen}, the one-step microwave-assisted method [(n-C4H9)4N]PF6, and are reported in Table 1. The electro- produced lower yields than a two-step procedure involving chemical and spectroelectrochemical characteristics of 2+ 4+ the initial synthesis of [Ru(pp)2(bpm)] , followed by its [{Ru(bpy)2}2(l-bpm)] (as a diastereoisomeric mixture)

Table 1 + 0 4+ Electrochemical data (in mV relative to the Fc /Fc couple) for the dinuclear complexes [{Ru(pp)2}2(l-BL)] {BL = bpm, dbneil} in 0.1 M [(n- a C4H9)4N]PF6/CH3CN b Complex DEox Eox2 Eox1 Ered1 Ered2 Ered3 Ered4 4+ c c,e meso-[{Ru(bpy)2}2(l-bpm)] 192 1384 1192 794 1466 1948 2368 4+ c c,e rac-[{Ru(bpy)2}2(l-bpm)] 188 1380 1192 792 1480 1928 2360 0 4+ c meso-[{Ru(5,5 -Me2bpy)2}2(l-bpm)] 188 1260 1072 824 1508 2032 2272 0 4+ c rac-[{Ru(5,5 -Me2bpy)2}2(l-bpm)] 176 1248 1072 824 1532 2032 2272 4+ c meso-[{Ru(Me4bpy)2}2(l-bpm)] 200 1208 1008 872 1460 2055 2315 4+ c rac-[{Ru(Me4bpy)2}2(l-bpm)] 192 1200 1008 880 1464 2056 2320 t 4+ c meso-[{Ru( Bu2bpy)2}2(l-bpm)] 192 1280 1088 816 1488 2008 2281 t 4+ c rac-[{Ru( Bu2bpy)2}2(l-bpm)] 184 1280 1088 816 1490 2015 2292 4+ d meso-[{Ru(Me2phen)2}2(l-bpm)] 192 1270 1078 816 1488 2008 2272 4+ d d rac-[{Ru(Me2phen)2}2(l-bpm)] 184 1266 1082 816 1512 2016 2264 4+ e meso-[{Ru(bpy)2}2(l-dbneil)] 180 1268 1088 544 945 1676 1972 4+ rac-[{Ru(bpy)2}2(l-dbneil)] 182 1270 1088 540 946 1668 1965 All potentials are reported ±3 mV. a All potentials are associated with one-electron electrochemical processes unless otherwise stated. b DEox = Eox2 Eox1. c Two-electron reduction process. d Process complicated by adsorption/desorption peaks. e Irreversible reduction process. D.M. DAlessandro, F.R. Keene / Chemical Physics 324 (2006) 8–25 15 have been detailed in a number of previous studies [30,71– Table 2 UV–visible–NIR spectral data of the reduced absorption spectra (e/m vs. m) 74,78–83]. n+ n+ The electrochemical characteristics of the diastereoiso- for the [Ru(bpy)2(BL)] and [{Ru(bpy)2}2(l-BL)] {BL = bpm, dbneil} 4+ in 0.02 M [(n-C4H9)4N]{B(C6F5)4}/CH3CN at +25 C mers of [{Ru(bpy)2}2(l-dbneil)] have been discussed pre- 1 1 viously [68,69,84]. Complex n+ mmax ± 10/cm {(e/m)max ± 0.0001/M } 4+ 0 bpm dbneil [69] The [{Ru(pp)2}2(l-bpm)] {pp = bpy, 5,5 -Me2bpy, t n+ Me4bpy, Me2phen and Bu2bpy} complexes are character- meso-[{Ru(bpy)2}2(l-BL)] 4 16775 (0.6142) 14290 (1.975) ized by two reversible one-electron redox processes corre- sh 18200 (0.4173) sh 22270 (1.131) sponding to successive oxidation of the metal centers, in 24270 (1.550) 23700 (1.591) addition to multiple reversible ligand-based reductions. In 5 5055 (0.1781) 4650 (0.2112) the cathodic region, the first two one-electron reduction 13540 (0.2463) 14045 (1.007) 16440 (0.2880) 23660 (1.005) processes are localized on the the bridging ligand (i.e., 24340 (0.6772) bpm0/ and bpm/2), due to the stronger p-acceptor nat- n+ ure of bpm relative to the terminal pp ligands [30,71–74,78– rac-[{Ru(bpy)2}2(l-BL)] 4 16730 (0.6143) 14290 (3.219) 83]. The subsequent reduction processes are localized on sh 18200 (0.3895) sh 22195 (1.813) the terminal pp ligands. 24230 (1.551) 23780 (2.590) The influence of the peripheral ligands on the electro- 5 5080 (0.1781) 4560 (0.2064) chemical characteristics of this series of substituted com- 13540 (0.2504) 14070 (1.660) plexes [{Ru(pp) } (l-bpm)]4+ is consistent with the 16395 (0.2952) 23070 (1.660) 2 2 24300 (0.6801) variation in the relative p-accepting abilities of the ligands [85]. For the diastereoisomers of a given complex, E and The NIR spectral data are indicated in bold type. ox1 sh, shoulder band. Eox2 shift cathodically as the peripheral ligands are varied t 0 through the series bpy > Bu2bpy Me2phen > 5,5 - Me2bpy > Me4bpy. Minor variations in DEox (DEox = Eox2 Eox1) are also evident across the series of complexes, and between the diastereoisomeric forms of the same complex. The potentials of the ligand-based reduction processes shift cathodically through the series bpy > Me2phen t 0 Bu2bpy > 5,5 -Me2bpy > Me4bpy. For a given complex, DEox is slightly greater for the meso compared with the rac diastereoisomeric form, however this difference is significant for the Me2phen and t Bu2bpy species only. The separation in the potentials between the metal-based oxidation processes permitted the generation of the mixed-valence forms of the complexes. The relative magnitudes of DEox suggest that the stabilities of the mixed-valence species are comparable for the series of complexes [30,86], although some caution must be exer- cised in the interpretation of DEox values [87]. The UV–visible–NIR spectral data for the un-oxidized (+4) and mixed-valence (+5) forms of the meso and rac n+ Fig. 4. UV–visible–NIR spectroelectrochemical progression for the oxi- diastereoisomers of [{Ru(bpy)2}2(l-bpm)] and [{Ru(b- 4+ dation reaction rac-[{Ru(bpy)2}2(l-bpm)] ! rac-[{Ru(bpy)2}2(l- n+ 1 5+ py)2}2(l-dbneil)] [87] over the range 3050–30000 cm bpm)] in 0.02 M [(n-C4H9)4N]{B(C6F5)4}/CH3CN at +25 C. The break are reported in Table 2, and the spectral progressions in the axis signifies that the spectra were obtained at different scan rates. accompanying the formation of the mixed-valence forms Inset. Overlay of IVCT bands for meso (—) and rac (----) diastereoisomers of the rac diastereoisomers are shown in Figs. 4 and 5, and the components obtained by Gaussian deconvolution for the meso form. respectively. The spectral features for the diastereoisomeric forms of the complexes are consistent with previous litera- 4+ 1 ture reports for [{Ru(bpy)2}2(l-bpm)] [30,71–74,88] (as a 16730 cm in the meso and rac diastereoisomers of 4+ 4+ diastereoisomeric mixture) and [{Ru(bpy)2}2(l-dbneil)] [{Ru(bpy)2}2(l-bpm)] , respectively, and at 14300 and [68,69]. 14290 cm1 in the meso and rac diastereoisomers of 4+ The UV–visible spectra over the region 10000– [{Ru(bpy)2}2(l-dbneil)] , respectively, are assigned as 30000 cm1 for the un-oxidized +4 forms of [{Ru- dp(RuII) ! p*(BL) transitions. These assignments are sup- 4+ (bpy)2}2(l-BL)] {BL = bpm, dbneil} are characterized ported by comparisons with the mononuclear complexes II 2+ 2+ by a combination of overlapping dp(Ru ) ! p*(BL) and [Ru(bpy)2(bpm)] [72,73,80] and [Ru(bpy)2(dbneil)] II 1 2+ dp(Ru ) ! p*(bpy) singlet metal-to-ligand ( MLCT) tran- [68] and the well-documented transitions in [Ru(bpy)3] sitions. The lowest energy transitions at 16775 and [89,90]. 16 D.M. DAlessandro, F.R. Keene / Chemical Physics 324 (2006) 8–25

fore appreciable net reduction of the 5+ to the 4+ ion occurred. Chemical decomposition was eliminated as a source of the instability of the mixed-valence species as the regeneration of the un-oxidized +4 species was achieved with >98% reversibility, and the decrease in the IVCT intensity (following the attainment of the maximum intensity) was consistent with the disproportion- ation of the mixed-valence species according to expected second-order kinetics [36,92,93].

3.4. Intervalence charge transfer

IVCT measurements for the complexes [{Ru(bpy)2}2(l- 5+ 5+ bpm)] and [{Ru(bpy)2}2(l-dbneil)] were performed in 0.02 M [(n-C4H9)4N]{B(C6F5)4}/CH3CN containing a uni- Fig. 5. UV–visible–NIR spectroelectrochemical progression for the oxi- form low concentration of the given diastereoisomer 4+ 3 dation reaction rac-[{Ru(bpy)2}2(l-dbneil)] ! rac-[{Ru(bpy)2}2(l- (0.40 10 M) to eliminate ion-pairing artefacts 5+ · dbneil)] in 0.02 M [(n-C4H9)4N]{B(C6F5)4}/CH3CN at +25 C. Inset. [20,21,57,94]. The {B(C6F5)4} anion is known to associate Overlay of IVCT bands for meso (—) and rac (----) diastereoisomers and the components obtained by Gaussian deconvolution for the meso form. weakly in comparison with PF6 , traditionally used for electrochemical measurements [63] . The NIR spectra of the dinuclear systems were scaled as e(m)/mdm [31,58] and Spectroelectrochemical generation of the mixed-valence deconvoluted by use of the software package GRAMS32. forms of meso- and rac-[{Ru(bpy) } (l-BL)]5+ (BL = bpm 2 2 The results of the band maxima, mmax, molar extinction and dbneil) revealed stable isosbestic points in the spectral coefficients, (e/m)max and bandwidths, Dm1/2, are summa- progressions, as shown in Figs. 4 and 5, respectively. The rized in Table 3. MLCT absorption bands decreased in intensity, and expe- The NIR band manifolds appear asymmetrical and rienced a slight red-shift following one-electron oxidation slightly narrower on the lower energy side with bandwidths to the mixed-valence species, with the appearance of a 1 at half-height (Dm1/2) of 2800 and 2620 cm for meso- and new band in the NIR region 3000–9000 cm1 (Figs. 4 5+ rac-[{Ru(bpy)2}2(l-bpm)] , respectively, and 1970 and and 5; Table 2). The NIR bands are absent in the spectrum 1 5+ 2050 cm for meso-andrac-[{Ru(bpy)2}2(l-dbneil)] , of the un-oxidized +4 species, and decrease in intensity as respectively. On the basis of a classical two-state model the potential is held at a value beyond that required for 0 the theoretical band-width at half-height, Dm1=2, is given generation of the +6 species. On this basis the bands at by [16RTln(2)m ]1/2 for a weakly coupled (Class II [95]) 1 max 4900 and 4890 cm in meso-andrac-[{Ru(bpy)2}2- system, where 16RTln(2) = 2310 cm1 at 298 K [96]. The 5+ 1 (l-bpm)] and at 4650 and 4560 cm in meso- and relatively narrow observed bandwidths suggest that the 5+ rac-[{Ru(bpy)2}2(l-dbneil)] are assigned as IVCT localized Class II description may be inappropriate for transitions. these systems [97]. The characterization of the +6 states of the complexes The parameter C provides a criterion for the degree of was not possible as the complete generation of the fully oxi- electronic coupling in the system [35]: dized forms could not be achieved reversibly. The energies III 1=2 of the Ru -based LMCT transitions in the dinuclear spe- C ¼ 1 ðDm1=2Þ=½16RT lnð2Þmmax cies could not be established, however a comparison with ¼ 1 ðDm Þ=ðDm0 Þ; ð4Þ 3+ 1=2 1=2 the mononuclear species [Ru(bpy)2(bpm)] [72,73,80] 3+ and [Ru(bpy)2(dbneil)] [68] suggests that such transi- where 0 < C < 0.5 for weak to moderate coupling (Class tions, if present, should occur in the visible region between II), C 0.5 at the Class II–III transition, and C > 0.5 for 10000 and 15000 cm1. This provides further support for strongly coupled (Class III) systems within the Robin and the assignment of the NIR bands as IVCT, rather than Day classification scheme [95]. For [{Ru(bpy)2}2(l- 5+ LMCT, transitions. bpm)] , C = 0.175 (meso) and 0.229 (rac) and for [{Ru- 5+ 5+ In previous work on [{Ru(bpy)2}2(l-bpm)] (as a dia- (bpy)2}2(l-dbneil)] , C = 0.40 (meso) and 0.37 (rac), stereoisomeric mixture) [30,78,91], an analysis of the which suggests that all systems lie between the fully local- IVCT characteristics was precluded as the mixed-valence ized (Class II) and delocalized (Class III) regimes, in the form was not sufficiently stable for a meaningful band- Class II–III transition region. width to be obtained. In the present case, a fast scanning Within the framework of the classical model, the asym- technique was employed over the wavelength range 3200– metric appearance of the bands is ascribed to the band 1 1 9200 cm (at a rate of 8000 cm /min), and permitted cut-off which occurs at hm =2Hab [35,95,97], where Hab is the generation of the mixed-valence species for a sufficient the electronic coupling parameter. When 0 < C < 0.5, Hab time to provide characterization of the IVCT bands be- is given by D.M. DAlessandro, F.R. Keene / Chemical Physics 324 (2006) 8–25 17

Table 3 Characteristics of the IVCT bands for the absorption spectra scaled as (e/m vs. m) for the dinuclear mixed-valence complexes in 0.02 M [(n- a,b C4H9)4N]{B(C6F5)4}/CH3CN at +25 C (parameters for overall envelope are shown in bold type: details of deconvoluted bands are in normal type) 1 1 1 0 1 1 Complex mmax ± 10/cm (e/m)max ± 0.0001/M Dm1/2 ± 10/cm Dm1=2/cm C Hab/cm 5+ meso-[{Ru(bpy)2}2(l-bpm)] 5055 0.1781 2800 3395 0.175 420 3615 0.0664 995 2870 0.347 110 5055 0.1779 2145 3395 0.368 365 7185 0.0193 1505 4050 0.372 145

5+ rac-[{Ru(bpy)2}2(l-bpm)] 5080 0.1781 2620 3400 0.229 405 3920 0.0479 710 2990 0.237 280 5090 0.1775 2145 3405 0.370 370 7160 0.0187 1520 4040 0.376 140

5+ meso-[{Ru(bpy)2}2(l-dbneil)] [69] 4650 0.2112 1970 3270 0.398 250 4570 0.1549 1595 3240 0.508 190 5530 0.07953 2390 3265 0.268 200

5+ rac-[{Ru(bpy)2}2(l-dbneil)] [69] 4560 0.2064 2050 3232 0.366 220 4495 0.1493 1465 3215 0.544 155 5515 0.08983 2290 3560 0.357 185 a 0 1/2 Dm1=2=[2310(mmax)] at 298 K. b ˚ ˚ Lower limits for Hab using rab = 7.9 A for dbneil [68] and 5.56 A for bpm [48]. These may diﬀer for the two diastereoisomers.

2 1=2 H ab ¼ 2:06 10 ðmmaxemaxDm1=2Þ =rab; ð5Þ sented in Table 4. The plot in Fig. 6 reveals the predicted linear trend for both diastereoiosmers in all solvents except where rab is the effective electron transfer distance. While AN. From the data (excluding AN), the following values rab is often equated with the through-space geometrical dis- were obtained for the slope and intercept: (meso) tance between the metal centers [98], the effective charge slope = 3545 ± 390 cm1 A˚ 1 and intercept = 3390 ± transfer distance is decreased relative to the geometric dis- 245 cm1 (R2 = 0.98); (rac) slope = 3690 ± 180 cm1 A˚ 1 tance as electronic coupling across the bridge increases, and and intercept = 3780 ± 115 cm1 (R2 = 0.99). The slopes this equation provides a lower limit only for Hab [98]. With of the plots are identical (within experimental error) for this caveat noted, Hab values for the meso and rac diastere- both diastereoisomers while the intercept for the rac form oisomers are shown in Table 3,withrab equated with the is marginally greater than the meso form. Minor differences approximate geometric metal–metal separation of 7.9 A˚ ˚ in the energies of the IVCT bands (mmax (meso rac)in for the dbneil-bridged species [68] and 5.56 A for the Table 4) are also apparent between the diastereoisomers bpm-bridged analogue [48]. across the series of solvents. The origin of the energy disparity in AN is attributed to 3.5. IVCT solvatochromism – the diastereoisomers of 5+ a specific solvent effect which overwhelms the dielectric [{Ru(bpy)2}2(l-bpm)] as probes for solvent continuum description. The clefts between the terminal reorganizational effects in IVCT bpy rings in the diastereoisomeric forms of [{Ru- (bpy) } (l-bpm)]5+ permit the AN molecules to approach The NIR spectra for the diastereoisomers of [{Ru- 2 2 5+ the metal centers more closely than is permitted by the the- (bpy)2}2(l-bpm)] were measured in the range of solvents oretical model: the clefts between the ligands at the two AN, PN, BN, iBN and BzN. The energies of the IVCT Ru(bpy)2 terminii (the ‘‘exterior clefts’’) are identical in band maxima (mmax) as a function of 1/Dop 1/Ds are pre- both diastereoisomeric forms while the dimensions of the

Table 4 5+ IVCT solvatochromism data of the reduced absorption spectra (e/m vs. m) for the diastereoisomers of [{Ru(bpy)2}2(l-bpm)] in 0.02 M [(n- a,b C4H9)4N]{B(C6F5)4}/solvent at +25 C; MLCT energies for the +4 states are also tabulated

Solvent 1/Dop 1/Ds meso rac

mmax ± Dm1/2 ± mMLCT(1) ± mmax ± Dm1/2 ± mMLCT(1) ± Dmmax (meso rac)± 10/cm1 10/cm1 10/cm1 10/cm1 10/cm1 10/cm1 10/cm1 AN 0.5127 5055 2800 16775 5080 2620 16775 25 PN 0.5011 5435 2710 16900 5470 2770 16900 35 iBN 0.4795 5420 3200 16850 5425 2952 16780 5 BN 0.4762 5395 2686 16800 5375 2860 16750 20 BzN 0.3897 5065 2570 16695 5095 2674 16660 30 a IVCT characteristics are reported as an average of triplicate experiments. b The absolute intensities of the IVCT bands, (e/m)max, are not tabulated as all spectra were normalized at the maximum intensity of the IVCT manifold. 18 D.M. DAlessandro, F.R. Keene / Chemical Physics 324 (2006) 8–25

linear solvent dependence of mmax on 1/Dop 1/Ds in PN, BN, iBN and BzN is surprising if all the solvents engage in some form of specific solvation. In polar solvents, 1/Dop 1/Ds, and the value of Dop in the first solvation shell may be similar to that in bulk solution. This would ex- plain the apparent success of the continuum theory, even in the presence of specific solvation effects for these solvents. The anomalous result in AN suggests that these molecules engage in a different form of specific interaction with the diastereoisomers compared with PN, BN, iBN and BzN. To address the issue of continuum vs. specific solvation, the IVCT characteristics of the diastereoisomeric forms of 5+ [{Ru(bpy)2}2(l-bpm)] were investigated as a function of solvent composition in a series of solvent mixtures containing varying mole fractions of AN (nAN) and PN

Fig. 6. mmax as a function of the solvent parameter 1/Dop 1/Ds for the (nPN =1 nAN)(Fig. 7 and Supplementary Table S1). for meso (—) and rac (----) diastereoisomeric forms of [{Ru(bpy)2}2(l- Fig. 7 reveals striking differences in the dependence of 5+ bpm)] in 0.02 M [(n-C4H9)4N]{B(C6F5)4}/solvent at +25 C. mmax on solvent composition for the two diastereoisomeric forms. Three regions are discernable. For 0 6 nAN 6 0.02, the rac diastereoisomer exhibited a 150 ± 10 cm1 red-shift clefts above and below the plane of the bridging ligand (the with the addition of two mole equivalents of AN while the ‘‘interior clefts’’) are distinguishable in the two forms, as meso form exhibited a 30 ± 10 cm1 red-shift over the same shown in Fig. 1. The minor differences between the energies range. The ratio of the number of moles of AN to [{Ru(b- 5+ of the IVCT bands for the diastereoismers suggest that spe- py)2}2(l-bpm)] is 2:1 at nAN = 0.02. The results suggest cific ‘‘stereochemically directed’’ interactions may be pres- that the specific solvent effect in dilute AN occurs within ent in all solvents due to solvent penetration within the the interior clefts, since the exterior clefts are identical in interior clefts. The red-shifts in mmax of 380 ± 10 and both diastereoisomeric forms. The specific effect for the 390 ± 10 cm1 between PN and AN for the meso and rac rac diastereoisomer may well correspond to the association diastereoisomers, respectively, are not expected in view of of two AN molecules, one within each identical cleft either the similar bulk dielectric properties of AN and PN, and side of the bpm plane. The closer distance of approach of the similarity of the structures which differ only in the pres- the AN molecules in the rac form gives rise to the larger ence of an additional methyl group for PN. Despite the magnitude of the specific effect. For 0.02 6 nAN 6 0.8 subtle structural differences between the two solvents, the (meso) and 0.2 6 nAN 6 0.6 (rac), mmax is relatively invari- results suggest that the relatively small AN molecules may penetrate the clefts and approach the metal centers more closely than is permitted for PN due to steric hindrance of that additional methyl group. Two issues arise in the attempt to rationalize these observations. Firstly, the red-shift of the IVCT energy in AN relative to PN is surprising since the general expecta- tion is that increased specific solvation induces an additional contribution to the reorganizational energy, and hence a blue-shift. Previous solvatochromism experiments for complexes based on –Ru(NH3)5 [2] have revealed correlations between the energies of the IVCT bands and empirical solvent basicity parameters such as DN due to specific solvent–ammine H-bonding. In the present case, solvent– solute interactions are likely to involve electrostatic interactions between the solvent molecules and pockets of electron density on the bpy and bpm ligands. Secondly, specific solvation effects will be present in addition to dielectric continuum solvation. In previous studies of complexes based on –Ru(NH3)5,–trans-Ru(NH3)4(py) and –Ru(bpy)(NH3)3 0 Fig. 7. mmax as a function of solvent composition for the meso (—) and rac with pyz, 4-cyanopyridine and 4,4 -bpy bridging ligands 5+ [19], m was fitted to a dual-parameter equation including (----) diastereoisomeric forms of [{Ru(bpy)2}2(l-bpm)] in AN and PN max mixtures containing 0.02 M [(n-C H ) N]{B(C F ) }/solvent at +25 C. both 1/D 1/D and DN. In the present case, the appar- 4 9 4 6 5 4 op s The inset shows mmax as a function of the absolute number of AN ent success of dielectric continuum theory in explaining the molecules (relative to one mole of complex). D.M. DAlessandro, F.R. Keene / Chemical Physics 324 (2006) 8–25 19 ant to solvent composition. In the rac form, the AN mole- l .Dl DE ¼ s ð6Þ cules located in the interior clefts block access to associa- s r3f ðÞ tion by additional solvent molecules, while the The 1/r3 dependence of DE is such that solvent molecules composition of molecules within the exterior clefts remains s located in closer proximity to the metal centers will domi- essentially constant. The specific association of solvent nate the specific solvent effect. The distance dependence molecules within the first solvation shell increases the effec- may account (in part) for the larger energy shifts in AN: tive radius and decreases the sensitivity of m to variation max due to their relatively small size, the discrete AN molecules in the solvent composition. The composition in the exterior can approach the metal centers more closely than the other clefts remains constant over the same range for the meso solvents of the series. The greater magnitude of the specific form, and the composition at the interior clefts also re- effect for the rac vs. the meso diastereoisomer at low AN mains relatively constant due to the larger orthogonal- concentration is believed to reflect the ability of the AN shaped cleft compared with the rac form. In the latter the molecules to associate more strongly within the clefts be- AN molecules associated within the relatively smaller inte- tween the metal centers. rior clefts are not accessible to interaction with solvent The dependence of DE on the vector dot product of l molecules in the bulk solvent. Given the more open nature s s and Dl gives rise to the orientation dependence of the sol- of the interior clefts in the meso diastereoisomer, the solvent molecules located within the clefts of the chromo- vent molecules associated within these clefts are more phore. All else being constant, |DE | is smallest when the accessible to solvent molecules in the outer solvation shells s dipole moment of the solvent molecule is oriented perpen- and the bulk solution. The increased solvent–solvent inter- dicularly to the plane of the bridging ligand and greatest actions may give rise to the consistently higher m for the max when oriented parallel to this plane. Assuming that the meso compared with the rac form. For 0.8 6 n 6 1.0 AN red-shift of 150 ± 10 cm1 in m corresponds to the for- (meso) and 0.6 6 n 6 1.0 (rac), m decreases for both max AN max mation of a 2:1 complex between AN and the dinuclear diastereoisomers. The dependence of m on solvent com- max cation for the rac diastereoisomer, the results suggest that position is greater for the meso diastereoiosmer, as quanti- each AN dipole is oriented with the nitrogen of AC„N fied by the slope of 2020 ± 130 cm1 per unit n (meso) AN directed inwards towards the clefts at an angle vs. 843 ± 60 cm1 per unit n (rac) over the range AN 0 6 h <90 to the Ru–Ru charge transfer axis. The small 0.8 6 n 6 1.0. At n = 0.2 (and n = 0.8), the ratios AN AN PN sizes of the AN molecules are such that they may associ- of AN and PN molecules to the mixed-valence complex ate sufficiently close to solvate both metal centers simulta- are 36:1 and 8.8:1, respectively. The sharp red-shift in m max neously. By comparison, the additional methyl group in with increasing concentration of AN occurs at higher n , AN PN may restrict the close approach of the solvent dipoles and more rapidly for the meso diastereoisomer. This form to the metal centers. Each PN dipole is thus oriented to- exhibits a greater sensitivity to solvent structure effects wards the metal center with the higher partial positive compared with the rac diastereoisomer, and these effects charge (d+), giving rise to a greater positive reorganiza- are more pronounced for solvent mixtures containing high tional contribution to m . For PN and the other mem- concentrations of AN [99]. The parameter m decreases max max bers of the nitrile series, the magnitudes of the specific more gradually for the rac diastereoisomer as the solvent effects are relatively smaller compared with AN, such that molecules associated within the interior clefts are relatively when they are superimposed on their respective contin- more restricted towards interactions with the bulk solvent, uum contributions, apparent conformity with the theoret- and hence to solvent structure effects. ical prediction is obtained. The results indicate that the specific solvent effect of dis- Clearly, a quantitative test of the applicability of Eq. (6) crete solvent molecules in the immediate vicinity of the depends on a realistic estimation of the dimensions of the complex dominate the solvent reorganizational energy. clefts (which is ambiguous due to their non-spherical nat- The diastereoisomers offer a detailed insight into the sol- ure). However, the qualitative predictions for the orienta- vent reorganizational contribution from specific solvent tion and distance dependence from Eq. (6) are molecules on k : the results support the general hypothesis o compatible with the experimental results. DE is thus super- that the distance of approach and orientation of the discrete s imposed on the 1/D 1/D continuum theory prediction solvent molecules dictate the solvent shifts. op s and can be regarded as an additional energy contribution Due to the specific nature of the solvent interactions, a to k . quantitative model for such effects must treat the solvent o on the molecular level as discrete entities. The specific sol- 3.6. Varying the selectivity of solvent association – influence vent effect (DEs), due to the presence of an oriented solvent molecule on a charge transfer transition, may be inferred of the bridging ligand from London dipole solvation theory [100,101] (Eq. (6)), where l is the ground state dipole moment of the solvent The IVCT solvatochromism properties in the diastereoi- s 5+ molecule, Dl is the dipole change due to charge transfer, r osomeric forms of [{Ru(bpy)2}2(l-dbneil)] (dbneil = is the cavity radius and f() is a function of the solvent dibenzoeilatin, Fig. 2) were examined to assess the effect dielectric properties. of increasing the dimensions of the interior clefts by 20 D.M. DAlessandro, F.R. Keene / Chemical Physics 324 (2006) 8–25

Table 5 5+ IVCT solvatochromism data of the reduced absorption spectra (e/m vs. m) for the diastereoisomers of [{Ru(bpy)2}2(l-dbneil)] in 0.02 M [(n- C4H9)4N]{B(C6F5)4}/solvent at +25 C

Solvent meso rac Dmmax (meso rac) ± 10/cm1 mmax ± (e/m)max ± Dm1/2 ± mMLCT(1) ± mmax ± (e/m)max ± Dm1/2 ± mMLCT(1) ± 10/cm1 0.0001/M1 10/cm1 10/cm1 10/cm1 0.0001/M1 10/cm1 10/cm1 AN 4650 0.2112 1970 14290 4560 0.2064 2050 14290 90 4570 0.1549 1595 4495 0.1493 1465 5530 0.07953 2390 5515 0.08983 2290 PN 4610 0.2081 1929 14335 4670 0.1515 1764 14335 60 BN 4590 0.1890 1884 14340 4585 0.0855 1948 14340 5 iBN 4610 0.1953 1871 14350 4600 0.1929 1917 14350 10 BzN 4600 0.1800 1880 14170 4590 0.1500 1880 14170 10 MLCT energies for the +4 states are also tabulated. Parameters for overall envelope are shown in bold type: details of deconvoluted bands (in AN only) are in normal type.

5+ Fig. 8. mmax as a function of the solvent parameter 1/Dop 1/Ds for the meso (—) and rac (----) diastereoisomers of [{Ru(bpy)2}2(l-bpm)] and 5+ [{Ru(bpy)2}2(l-dbneil)] in 0.02 M [(n-C4H9)4N]{B(C6F5)4}/solvent at +25 C. comparison with the analogous diastereoisomers of sion for the dbneil-bridged diastereoisomers (Fig. 9), but 5+ [{Ru(bpy)2}2(l-bpm)] . The variation of the energies of larger than for a given diastereoisomer compared with their the IVCT band maxima (mmax) as a function of 1/Dop bpm-bridged analogues. Since the exterior clefts are identi- 5+ 1/Ds for [{Ru(bpy)2}2(l-dbneil)] are presented in Table 5 and Fig. 8. The complete NIR–UV–visible spectral data for the un-oxidized (+4) and mixed-valence (+5) n+ forms of [{Ru(bpy)2}2(l-dbneil)] have been published previously [69]. The energies of the IVCT bands of [{Ru(bpy)2}2(l- dbneil)] are essentially solvent-independent for both diastereoisomers, as shown in the plot of mmax as a function of 1/Dop 1/Ds (Fig. 8). The decreased slope of the solvatochromism plots compared with bpm-bridged analogues contradicts the prediction from Eq. (2) that ko increases as the intra-metal distance is increased from 5.56 A˚ (mea- 5+ Fig. 9. Chem 3D representations of the diastereoisomeric forms of sured for meso-[{Ru(Me2 bpy)2}2(l-bpm)] [48]) for the n+ ˚ [{Ru(bpy)2}2(l-dbneil)] illustrating the subtle variation in the dimen- bpm-bridged systems to 7.9 A [68] for the dbneil-bridged sions of the interior clefts above and below the plane of the bridging systems (at fixed a). The interior clefts are of similar dimen- ligand. Hydrogen atoms are omitted for clarity. D.M. DAlessandro, F.R. Keene / Chemical Physics 324 (2006) 8–25 21 cal for the diastereoisomers of both complexes, the results suggest that solvation in the interior clefts is more important than solvation about the entire dimer. In particular, the solvent reorganizational contribution to the IVCT energy is greater for the bpm-bridged diastereoisomers which contain the relatively smaller interior clefts. Alternatively, the dbneil-bridged diastereoisomers may exhibit relatively greater delocalization than their bpm- bridged analogues due to the extensive aromatic framework of the bridging ligand in the former case. The slope of the solvatochromism plot is given by e2(1/a 1/d) according to Eq. (2), however the equation is frequently ex- pressed in terms of the effective amount of charge transferred, and the slope is given instead as (De)2(1/a 1/d). Since the effective amount of charge transferred is reduced from unit charge transfer by delocalization, this explana- tion may provide a qualitative rationale for the lesser slope for the dbneil-bridged systems. The data for the IVCT parameters in Table 3 do indeed suggest that the latter are more delocalized as the bands are narrower and more intense. The solvent dependence of mmax for the dbneil diastereoisomers (Fig. 8) arises from a superposition of the solvent reorganizational contributions due to the continuum and specific solvation effects. A comparison of the difference in mmax between the diastereoisomers {Dmmax (meso rac)} 5+ 5+ for [{Ru(bpy)2}2(l-dbneil)] and [{Ru(bpy)2}2(l-bpm)] reveals that the differences in the specific solvent interactions between the diastereoisomeric forms are present for both complexes in each solvent. In particular, the difference in the specific solvent interaction with the diastereoisomers Fig. 10. mmax as a function of the solvent parameter 1/Dop 1/Ds for the of the same complex in AN and PN is greater for the dbneil 5+ meso and rac diastereoisomers of [{Ru(pp)2}2(l-bpm)] {pp = bpy (—), 0 t ...... complexes. This reflects the larger cleft available for solvent 5,5 -Me2bpy (– – –), Me4bpy (----), Me2phen (–Æ–Æ–) and Bu2bpy ( )} penetration relative to the bpm-bridged species. in 0.02 M [(n-C4H9)4N]{B(C6F5)4}/solvent at +25 C. Error bars are omitted for clarity. 3.7. Varying the selectivity of solvent association – influence of the terminal ligands tional contributions of specific solvent molecules to the to- The final investigation involved the systematic modifica- tal energy of the IVCT transition. The energies of the IVCT tion of the dimensions of the interior and exterior clefts by bands for the substituted derivatives exhibit a lesser but the judicious positioning of substituents on the terminal scattered dependence on 1/Dop 1/Ds (Fig. 10) relative 5+ ligands in the series [{Ru(pp)2}2(l-bpm)] , where to the linear dependence of mmax on 1/Dop 1/Ds for the 0 0 0 0 5+ pp = 5,5 -dimethyl-2,2 -bipyridine (5,5 -Me2bpy), 4,4 ,5, diastereoisomers of [{Ru(bpy)2}2(l-bpm)] in PN, BN, i 0 0 5 -tetramethyl-2,2 -bipyridine (Me4bpy), 2,9-dimethyl- BN and BzN. Negligible slopes are obtained for all com- 0 0 1,10-phenanthroline (Me2phen) and 4,4 -di-tert-butyl-2,2 - plexes (neglecting the results in AN). The striking disconti- t bipyridine ( Bu2bpy), shown in Fig. 3. nuity between PN and AN in the plot of mmax vs. 5+ The results for the energies of the IVCT band maxima as 1/Dop 1/Ds for the [{Ru(bpy)2}2(l- bpm)] diastereoiso- a function of 1/Dop 1/Ds for the diastereoisomeric forms mers is also evident for the substituted derivatives, with the 5+ 0 of the series [{Ru(pp)2}2(l-bpm)] {pp = 5,5 -Me2bpy, notable exceptions being the diastereoisomers of [{Ru- t t 5+ Me4bpy, Me2phen and Bu2bpy} in a homologous series ( Bu2bpy)2}2(l-bpm)] . The observation may be rational- of nitrile solvents are reported Supplementary Table S2 ized by the steric restriction of the bulky tert-butyl and shown in Fig. 10. substituents to access of solvent molecules to both the inte- Specific solvation effects clearly dominate the magnitude rior and exterior clefts. However, minor differences in the of ko, and the consideration of the subtle and systematic IVCT energies are evident between the two diastereoiso- structural variation between the different complexes, and meric forms in all solvents, as quantified by Dmmax (me- between the diastereoisomeric forms of the same complex, so rac) – provided in Supplementary Table S2 and provide insights into the detailed nature of the reorganiza- Supplementary Figure S1 – which represents differences 22 D.M. DAlessandro, F.R. Keene / Chemical Physics 324 (2006) 8–25

5+ 0 t Fig. 11. Chem 3D representations of [{Ru(pp)2}2(l-bpm)] {pp = 5,5 -Me2bpy, Me4bpy, Me2phen, Bu2bpy} illustrating the subtle variation in the nature of the clefts above and below the plane of the bridging ligand. Hydrogen atoms are omitted for clarity. in the specific solvent interactions at the interior clefts. The from directly above the interior clefts. For both complexes, rac diastereoisomer exhibits a negligible solvent shift the methyl substituents at the 5,50-positions restrict solvent (13 ± 10 cm1) between AN and PN due to the smaller access at the convergence of the terminal ligands in the and less solvent-accessible interior cleft compared with meso form, while the dimensions of the clefts in the rac the meso form, where the shift is 90 ± 10 cm1. If the con- form are comparable to those of the unsubstituted bpy tribution of specific effects are negligible for rac-[{Ru- complex. Despite the presence of the methyl substituents, t 5+ ( Bu2bpy)2}2(l-bpm)] , then Dmmax (meso rac) provides the magnitude of the specific solvent effect due to AN is a quantitative measure of the specific solvent effect at the maintained for both diastereoisomers of both complexes. interior cleft, and hence provides a clear partition between The difference between the diastereoisomers {Dmmax the magnitudes of the continuum and specific solvation ef- (meso rac)} is more pronounced in AN for the Me4bpy fects to ko. derivatives compared with all the substituted complexes, The subtle structural differences between the various which suggests that substitution at the 4,40-positions of 5+ [{Ru(pp)2}2(l-bpm)] complexes provide further opportu- the terminal bpy-type ligands results in the most pro- nities to quantitatively assess stereochemically directed spe- nounced difference in the specific interaction of the AN cific solvent effects. The diastereoisomers of the respective molecules at the interior clefts. For a given diastereoiso- complexes are shown in Fig. 11. mer, the energies of the IVCT transitions in PN, BN, i 0 BN and BzN are greater for the 5,5 -Me2bpy derivative, 0 5+ 3.7.1. [{Ru(5,5 -Me2bpy)2}2(l-bpm)] and which may reflect the smaller a (and hence larger contin- 5+ [{Ru(Me4bpy)2}2(l-bpm)] uum contribution, according to Eq. (2)) for the diastereoi- 0 5+ The dimensions of the interior and exterior clefts be- somers of [{Ru(5,5 -Me2bpy)2}2(l-bpm)] compared with 5+ tween the same diastereoiomeric forms of these two com- the corresponding forms of [{Ru(Me4bpy)2}2(l-bpm)] . plexes differ only with respect to the additional methyl For both diastereoiosomers of [{Ru(Me4bpy)2}2(l- 0 5+ substituents at the 4,4 -positions in [{Ru(Me4bpy)2}2(l- bpm)] , mmax remains essentially constant over the series bpm)]5+, which may hinder the access of solvent molecules PN, BN, iBN and BzN compared with the relatively D.M. DAlessandro, F.R. Keene / Chemical Physics 324 (2006) 8–25 23 greater solvent effects for the diastereoisomers of [{Ru- scribed previously for the diastereoisomers of the latter 0 5+ (5,5 -Me2bpy)2}2(l-bpm)] . The results indicate that the group of complexes in AN is absent as the tert-butyl sub- specific solvent interactions for the diastereoisomeric forms stituents block access to the interior clefts for the smallest 0 of the 5,5 -Me2bpy derivative may occur via solvent pene- solvent of the nitrile series. Small differences in mmax are tration from directly above or below the plane of the bridg- evident between the diastereoisomeric forms of [{Ru- t 5+ ing ligand, since the only difference between the same ( Bu2bpy)2}2(l-bpm)] as the solvent molecules may diastereoisomer for the two complexes is substitution at approach the interior clefts more readily in the meso diaste- 0 the 4,4 -positions of the Me2bpy rings. When tert-butyl reoisomer compared with the rac form. substituents occupy these positions, solvent access is se- The solvent dependence of the IVCT energies for the 5+ verely restricted, even for the relatively small AN diastereoisomers of the series [{Ru(pp)2}2(l-bpm)] 0 t molecules. {pp = 5,5 -Me2bpy, Me4bpy, Me2phen and Bu2bpy} are consistent with the superposition of continuum and specific 5+ 3.7.2. [{Ru(Me2phen)2}2(l-bpm)] solvation effects, in which the magnitude of the latter may The methyl substituents at the 2,9-positions of 1,10-phe- dominate the ko contribution to mmax. The subtle variations nanthroline (phen) induce a steric crowding close to the in the dimensions of the interior clefts between the diaste- metal centers which should restrict the solvent access in reoisomeric forms of the same complex, and between both the immediate vicinity of the metal centers to a greater ex- the interior and exterior clefts for the same diastereoisomer tent than substitution at the 4,40- and/or 5,50-positions of over the series of complexes, give rise to a marked distance the bpy-based ligands. In the meso diastereoisomer, the and orientation dependence of the specific effects on ko. methyl substituents induce a crowding at the convergence The solvent dependence of the IVCT energies for the dia- 5+ of the terminal rings on either side of the bridging ligand stereoisomers of [{Ru(bpy)2}2(l-bpm)] exhibits the theo- plane, and the size of the orthogonal-shaped interior retically predicted linear dependence on the dielectric cavities are comparable to those in meso-[{Ru(pp)2}2(l- parameter 1/Dop 1/Ds for all solvents except AN. The 5+ 0 bpm)] {pp = bpy, 5,5 -Me2bpy and Me4bpy}. The simi- striking discontinuity for AN arises from penetration of lar nature of the interior clefts for the complexes would the small AN molecules within the clefts between the termi- account for the similar nature of the specific solvent effect nal bpy rings. This specific interaction is also evident for 5+ observed in AN. The magnitude of the specific solvent the substituted variants [{Ru(pp)2}2(l-bpm)] 0 effect between AN and PN for rac-[{Ru(Me2phen)2}2(l- {pp = 5,5 -Me2bpy, Me4bpy, Me2phen}, however the effect 5+ bpm)] is comparable to that for rac-[{Ru(bpy)2}2(l- is absent when solvent penetration to the clefts is restricted 5+ t bpm)] , which suggests that the AN and PN dipoles by bulky tert-butyl substituents in [{Ru( Bu2bpy)2}2(l- may assume similar orientations and distances of approach bpm)]5+. The comparable magnitudes of the specific effects in both complexes. for both the unsubstituted bpy, and methyl-substituted 0 Substitution at the 2,9-positions gives rise to a dramatic {pp = 5,5 -Me2bpy, Me4bpy, Me2phen} complexes, sug- difference in mmax between the meso and rac diastereoiso- gest that the AN dipoles must assume a similar orientation i 1 mers in BN, where Dmmax (meso rac) is 180 ± 10 cm . to the bpm plane in each case. The red-shift of mmax be- The difference is striking in view of the subtle structural dif- tween PN and AN indicates that the AN molecules are ference between iBN, PN and BN. The additional steric associated within the interior clefts so as to solvate both bulk provided by the branched position of the methyl metal centers simultaneously, as discussed previously for i 5+ group in BN compared with PN and BN may be sufficient the [{Ru(bpy)2}2(l-bpm)] diastereoisomers. By compari- to inhibit penetration of these solvent molecules within the son, the slightly larger PN molecules are oriented towards interior clefts of the rac form, compared with the relatively the metal center with the higher partial positive charge more open clefts in the meso diastereoisomer. The magni- (d+), thus giving rise to a greater positive reorganizational tude of the difference between the diastereoisomers is sig- contribution to mmax. The nature of the specific solvent 0 nificantly greater than that observed for the 5,5 -Me2bpy interactions is markedly dependent on the position of the and Me4bpy derivatives. This suggests that the presence methyl substituents on the terminal ligands, and the size of methyl substituents at the 2,9-positions in Me2phen of the solvent molecules. The difference in the IVCT ener- gives rise to the greatest stereochemically induced differen- gies between the meso and rac forms is largest for 5+ tiation between the diastereoisomeric forms. [{Ru(Me2phen)2}2(l-bpm)] and provides the most pronounced example of differential stereochemically directed t 5+ 3.7.3. [{Ru( Bu2bpy)2}2(l-bpm)] solvent effects on ko. The tert-butyl substituents at the 4,40-positions of the t t 5+ terminal Bu2bpy ligands in [{Ru( Bu2bpy)2}2(l-bpm)] 4. Conclusions provide significantly greater steric hindrance to solvent penetration into the internal (and to a lesser extent, exter- IVCT solvatochromism studies on the meso and rac dia- 5+ nal) clefts relative to the diastereoisomers of stereoisomers of [{Ru(bpy)2}2(l-bpm)] in a homologous 5+ 0 [{Ru(pp)2}2(l-bpm)] {pp = bpy, 5,5 -Me2bpy, Me4bpy, series of nitrile solvents reveal that stereochemically direc- Me2phen}. Most notably, the specific solvent effect de- ted specific solvent effects in the first solvation shell domi- 24 D.M. DAlessandro, F.R. Keene / Chemical Physics 324 (2006) 8–25 nate the outer sphere contribution to the reorganizational [15] B.S. Brunschwig, S. Ehrenson, N. Sutin, J. Phys. Chem. 90 (1986) energy for intramolecular electron transfer. Solvent pro- 3657. portion experiments in AN/PN solvent mixtures demon- [16] S.F. Nelsen, R.F. Ismagilov, J. Phys. Chem. A 103 (1999) 5373. [17] R.A. Marcus, J. Phys. Chem. B 102 (1998) 10071. strate that the magnitude and direction of the specific [18] P. Pereztejeda, P. Neto-Ponce, F. Sańchez, J. Chem. Soc., Dalton effect is dependent on the relative abilities of discrete sol- Trans. (2001) 1686. vent molecules to penetrate the clefts between the planes [19] K.W. Lau, A.M.H. Hu, M.H.J. Yen, E.Y. Fung, S. Grzybicki, R. of the terminal polypyridyl ligands. In particular, the spe- Matamoros, J.C. Curtis, Inorg. Chim. Acta 226 (1994) 137. cific effects are dependent on the dimensionality of the [20] N.A. Lewis, Y.S. Obeng, W.L. Purcell, Inorg. Chem. 28 (1989) 3796. [21] N.A. Lewis, Y.S. Obeng, J. Am. Chem. Soc. 110 (1988) 2306. clefts, and the number, size, orientation and location of [22] J.T. Hupp, Y. Dong, Inorg. Chem. 33 (1994) 4421. the solvent dipoles within the interior and exterior clefts. [23] A. Ferretti, A. Lami, G. Villani, Inorg. Chem. 37 (1998) 4460. IVCT solvatochromism studies on the diastereoisomeric [24] J.T. Hupp, G.A. Neyhart, T.J. Meyer, E.M. Kober, J. Phys. Chem. 5+ 96 (1992) 10820. forms of [{Ru(bpy)2}2(l-dbneil)] and [{Ru(pp)2}2(l- bpm)]5+ {pp = 5,50-Me bpy, Me bpy, Me phen, tBu bpy} [25] J.T. Hupp, Y.H. Dong, J. Am. Chem. Soc. 115 (1993) 6428. 2 4 2 2 [26] Y. Dong, J.T. Hupp, Inorg. Chem. 31 (1992) 3322. reveal that the subtle and systematic changes in the nature [27] D. Catterjee, H.C. Bajaj, A. Das, Inorg. Chim. Acta 224 (1994) 189. of the clefts by the variation of the bridging ligand, and the [28] D.M. DAlessandro, L.S. Kelso, F.R. Keene, Inorg. Chem. 40 (2001) judicious positioning of substituents on the terminal 6841. ligands profoundly influence the magnitude of the reorga- [29] R. de la Rossa, P.J. Chang, F. Salaymeh, J.C. Curtis, Inorg. Chem. nizational energy contribution to the electron transfer 24 (1985) 4229. [30] K.A. Goldsby, T.J. Meyer, Inorg. Chem. 23 (1984) 3002. barrier. [31] N.S. Hush, Prog. Inorg. Chem. 8 (1967) 391. The present study provides an experimental platform [32] N.S. Hush, Electrochim. Acta 13 (1968) 1005. which addresses the paucity of experimental data available [33] R.A. Marcus, J. Chem. Phys. 26 (1957) 867. to guide and test developing solvation models which link [34] R.A. Marcus, J. Chem. Phys. 24 (1956) 966. the dynamics of individual solvent molecules and the Mar- [35] B.S. Brunschwig, C. Creutz, N. Sutin, Chem. Soc. Rev. 31 (2002) 168. cus–Hush theory of electron transfer [2,8,40,101,102]. [36] C. Creutz, Prog. Inorg. Chem. 30 (1983) 1. [37] R.J. Crutchley, Adv. Inorg. Chem. 41 (1994) 273. Acknowledgments [38] J.T. Hupp, T.J. Meyer, Inorg. Chem. 26 (1987) 2332. [39] J.T. Hupp, T.J. Meyer, J. Phys. Chem. 91 (1987) 1001. [40] P.F. Barbara, T.J. Meyer, M.A. Ratner, J. Phys. Chem. 100 (1996) We thank Professor Noel Hush and Dr. Jeff Reimers for 13148. helpful discussions on this work, and gratefully acknowl- [41] J.C. Curtis, B.P. Sullivan, T.J. Meyer, Inorg. Chem. 22 (1983) 224. edge the financial support of the Australian Research [42] V. Gutmann, Electrochim. Acta 21 (1976) 661. Council. [43] M.H. Chou, C. Creutz, N. Sutin, Inorg. Chem. 31 (1992) 2318. [44] C. Creutz, M.H. Chou, Inorg. Chem. 26 (1987) 2995. [45] G.A. Neyhart, C.J. Timpson, W.D. Bates, T.J. Meyer, J. Am. Chem. Appendix A. Supplementary data Soc. 118 (1996) 3730. [46] F.R. Keene, Coord. Chem. Rev. 166 (1997) 122. [47] F.R. Keene, Chem. Soc. Rev. 27 (1998) 185. Supplementary data associated with this article can be [48] N.C. Fletcher, P.C. Junk, D.A. Reitsma, F.R. Keene, J. Chem. Soc., found, in the online version, at doi:10.1016/j.chemphys. Dalton Trans. (1998) 133. 2005.09.016. [49] N.C. Fletcher, F.R. Keene, J. Chem. Soc., Dalton Trans. (1999) 683. [50] D. Perrin, W. Amarego, Purification of Laboratory Chemicals, Pergamon Press, Oxford, 1992, pp. 192–193. References [51] B.T. Patterson, F.R. Keene, Inorg. Chem. 37 (1998) 645. [52] L. Della Ciana, W.J. Dressick, D. Sandrini, M. Maestri, M. Ciano, [1] C. Creutz, H. Taube, J. Am. Chem. Soc. 95 (1973) 1086. Inorg. Chem. 29 (1990) 2792. [2] P. Chen, T.J. Meyer, Chem. Rev. 98 (1998) 1439. [53] N.G. Connelly, W.E. Geiger, Chem. Rev. 96 (1996) 877. [3] S.F. Nelsen, D.A. Trieber II, R.F. Ismagilov, Y. Teki, J. Am. Chem. [54] C.M. Duff, G.A. Heath, Inorg. Chem. 30 (1991) 2528. Soc. 123 (2001) 5684. [55] D.M. DAlessandro, F.R. Keene, Chem. Eur. J. 11 (2005) 3679. [4] J.T. Hupp, M.J. Weaver, Inorg. Chem. 23 (1984) 3639. [56] M.D. Lowery, W.S. Hammack, H.G. Drickamer, D.N. Hendrick- [5] J. Katriel, M.A. Ratner, J. Phys. Chem. 93 (1989) 5065. son, J. Am. Chem. Soc. 109 (1987) 8019. [6] R.S. Drago, D.E. Richardson, J.E. George, Inorg. Chem. 36 (1997) [57] R.L. Blackbourn, J.T. Hupp, J. Phys. Chem. 94 (1990) 1788. 25. [58] J.R. Reimers, N.S. Hush, Inorg. Chem. 29 (1990) 3686. [7] E.R. Barthel, I.B. Martini, B.J. Schwartz, J. Phys. Chem. B 2001 [59] J.-P. Launay, Chem. Soc. Rev. 30 (2001) 386. (2001) 12230. [60] D.M.P. Mingos, D.R.D. Baghurst, Chem. Soc. Rev. 20 (1991) 1. [8] D.V. Matyushov, R. Schmid, J. Phys. Chem. 98 (1994) 5152. [61] E.Z. Jandrasics, Ph.D. Thesis, University of Fribourg (Switzerland), [9] B.P. Sullivan, J.C. Curtis, E.M. Kober, T.J. Meyer, Nouv. J. Chim. Fribourg, 1995. 4 (1980) 643. [62] E.Z. Jandrasics, F.R. Keene, J. Chem. Soc., Dalton Trans. (1997) [10] J.A. Roberts, J.T. Hupp, Inorg. Chem. 31 (1992) 157. 153. [11] R.L. Blackbourn, J.T. Hupp, J. Phys. Chem. 92 (1988) 2817. [63] R. LeSuer, W.E. Geiger, Angew. Chem. Int. Ed. 39 (2000) 248. [12] R.L. Blackbourn, J.T. Hupp, Inorg. Chem. 28 (1989) 3786. [64] I.P. Evans, A. Spencer, G. Wilkinson, J. Chem. Soc., Dalton Trans. [13] J.T. Hupp, J. Weydert, Inorg. Chem. 26 (1987) 2657. (1973) 204. [14] K.S. Ennix, P.T. McMahon, R. de la Rosa, J.C. Curtis, Inorg. [65] T. Togano, N. Nagao, M. Tsuchida, H. Kumakura, K. Hisamatsu, Chem. 26 (1987) 2660. F.S. Howell, M. Mukaida, Inorg. Chim. Acta 195 (1992) 221. D.M. DAlessandro, F.R. Keene / Chemical Physics 324 (2006) 8–25 25

[66] P.A. Anderson, R.F. Anderson, M. Furue, P.C. Junk, F.R. Keene, [84] D. Gut, I. Goldberg, M. Kol, Inorg. Chem. 42 (2003) 3483. B.T. Patterson, B.D. Yeomans, Inorg. Chem. 39 (2000) 2721. [85] G. Giuﬀrida, S. Campagna, Coord. Chem. Rev. 135 (1994) 517. [67] B.P. Sullivan, D.J. Salmon, T.J. Meyer, Inorg. Chem. 17 (1978) [86] E.M. Kober, K.A. Goldsby, D.N.S. Narayana, T.J. Meyer, J. Am. 3334. Chem. Soc. 105 (1983) 4303. [68] S.D. Bergman, I. Goldberg, A. Barbieri, F. Barigelletti, M. Kol, [87] D.M. DAlessandro, F.R. Keene, Dalton Trans. (2004) 3950. Inorg. Chem. 43 (2004) 2355. [88] G.N.A. Nallas, S.W. Jones, K.J. Brewer, Inorg. Chem. 35 (1996) [69] D.M. DAlessandro, F.R. Keene, S.D. Bergman, M. Kol, Dalton 6974. Trans. (2005) 332. [89] F.E. Lytle, D.M. Hercules, J. Am. Chem. Soc. 91 (1969) 253. [70] X. Hua, Ph.D. Thesis, University of Fribourg (Switzerland), [90] A. Juris, S. Barigelletti, S. Campagna, V. Balzani, P. Belser, A. von Fribourg, 1993. Zelewsky, Coord. Chem. Rev. 84 (1988) 85. [71] M. Hunziker, A. Ludi, J. Am. Chem. Soc. 99 (1977) 7370. [91] F. Baumann, W. Kaim, M.C. Posse, N.E. Katz, Inorg. Chem. 37 [72] E.V. Dose, L.J. Wilson, Inorg. Chem. 17 (1978) 2660. (1998) 658. [73] D.P. Rillema, K.B. Mack, Inorg. Chem. 21 (1982) 3849. [92] M.D. Ward, Chem. Soc. Rev. 24 (1995) 121. [74] M. Krejcik, A.A. Vlcek, Inorg. Chem. 31 (1992) 2390. [93] K. Kalyanasundaram, M.K. Nazeeruddin, Inorg. Chim. Acta 226 [75] X. Hua, A. von Zelewsky, Inorg. Chem. 30 (1991) 3796. (1994) 213. [76] X. Hua, A. von Zelewsky, Inorg. Chem. 34 (1995) 5791. [94] R.L. Blackbourn, J.T. Hupp, Chem. Phys. Lett. 150 (1988) 399. [77] D.A. Reitsma, F.R. Keene, J. Chem. Soc., Dalton Trans. (1993) [95] M.B. Robin, P. Day, Adv. Inorg. Chem. Radiochem. 10 (1967) 247. 2859. [96] N.S. Hush, Coord. Chem. Rev. 64 (1985) 135. [78] R.A. Pavinato, J.A. Walk, M.E. McGuire, Inorg. Chem. 32 (1993) [97] S.F. Nelsen, Chem. Eur. J. 6 (2000) 581. 4982. [98] K.D. Demadis, C.M. Hartshorn, T.J. Meyer, Chem. Rev. 101 (2001) [79] R.R. Ruminski, J.D. Petersen, Inorg. Chem. 21 (1982) 3706. 2655. [80] D.P. Rillema, G. Allen, T.J. Meyer, D. Conrad, Inorg. Chem. 22 [99] J.R. Reimers, L.E. Hall, J. Am. Chem. Soc. 121 (1999) 3730. (1983) 1617. [100] N.S. Hush, J.R. Reimers, Coord. Chem. Rev. 177 (1998) 37. [81] S. Ernst, V. Kasack, W. Kaim, Inorg. Chem. 27 (1988) 1146. [101] N.S. Hush, J.R. Reimers, Chem. Rev. 100 (2000) 775. [82] S.D. Ernst, W. Kaim, Inorg. Chem. 28 (1989) 1520. [102] D.W. Small, D.V. Matyushov, G.A. Voth, J. Am. Chem. Soc. 125 [83] W. Kaim, S. Kohlmann, Inorg. Chem. 26 (1987) 68. (2003) 7470.