Unusually Fast Electron and Anion Transport Processes Observed in

School of Chemistry, Dublin Institute of Technology, KeVin Street, Dublin 8, Ireland ... School of Chemical Sciences, Dublin City UniVersity, GlasneVi...
0 downloads 0 Views 151KB Size
J. Phys. Chem. B 2000, 104, 1977-1983

1977

Unusually Fast Electron and Anion Transport Processes Observed in the Oxidation of “Electrochemically Open” Microcrystalline [{M(bipy)2}{M′(bipy)2}(µ-L)](PF6)2 Complexes (M, M′ ) Ru, Os; bipy ) 2,2′-Bipyridyl; L ) 1,4-Dihydroxy-2,5-bis(pyrazol-1-yl)benzene Dianion) at a Solid-Electrode-Aqueous Electrolyte Interface Alan M. Bond*,† Department of Chemistry, Monash UniVersity, Clayton, Victoria 3168, Australia

Frank Marken,* Christopher T. Williams, and David A. Beattie Physical & Theoretical Chemistry Laboratory, UniVersity of Oxford, South Parks Road, Oxford OX1 3QZ, England

Tia E. Keyes* School of Chemistry, Dublin Institute of Technology, KeVin Street, Dublin 8, Ireland

Robert J. Forster and Johannes G. Vos School of Chemical Sciences, Dublin City UniVersity, GlasneVin, Dublin 9, Ireland ReceiVed: July 26, 1999; In Final Form: October 30, 1999

The oxidation of a series of bridged dinuclear metal complexes [{M(bipy)2}{M′(bipy)2}(µ-L)]2+ (M, M′ ) Ru, Os; bipy ) 2,2′-bipyridyl; L ) 1,4-dihydroxy-2,5-bis(pyrazol-1′-yl)benzene dianion) in microcrystalline solid form has been studied at the basal plane pyrolytic graphite electrode-aqueous electrolyte interface. The solid materials undergo unusually rapid and essentially exhaustive electrolysis even under fast scan rate conditions of cyclic voltammetry via two one-electron oxidation charge-transfer steps: [{M(bipy)2}{M′(bipy)2}(µ-L)]X2(solid) + X-(solution) a [{M(bipy)2}{M′(bipy)2}(µ-L)]X3(solid) + e- (step 1); [{M(bipy)2}{M′(bipy)2}(µ-L)]X3(solid) + X-(solution) a [{M(bipy)2}{M′(bipy)2}(µ-L)]X4(solid) + e- (step 2). To maintain charge neutrality, the solid-state charge-transfer processes are coupled to rapid insertion/expulsion of anions from/to the aqueous electrolyte solution phase (X- ) PF6-, ClO4-, SCN-, or NO3-). The conclusion is reached that “electrochemically open” solid-state structural features are responsible for the uncommonly fast electron transfer and anion charge neutralization processes, which are of relevance to charge storage and photochromic devices and which proceed without nucleation and redistribution processes frequently identified in other systems. Thus, a description can be based on an interacting thin layer model with a “Donnan” type potential term for the anion dependence of the reversible potential. In situ spectroelectrochemical measurements (controlled potential Raman spectroscopy) and ex situ scanning electron microscopy studies yield detailed complementary information concerning the solid-state aspects of the redox transformations, which are localized on the dioxolene bridge and correspond to reversible hydroquinone, semiquinone, and quinone interconversions.

Introduction The study of the redox reactions of solid materials by voltammetric techniques has found a wide range of applications in fields of research concerned with energy storage1 and electrochromic devices.2 Significant developments have been covered in recent reviews.3,4 In particular, the possibility to mechanically attach materials, as synthesized, to suitable electrode surfaces without sophisticated deposition methods has widened the horizons of conventional voltammetric techniques5-7 and enabled the study of new phenomena, such as the “matrix” stabilization of reactive solution phase intermediates.8 However, even with these advances there still remains a considerable need to broaden the knowledge base available on reactions that take † Telephone: (61)(3) 9905 1338. Fax: (61)(3) 9905 4597. E-mail: [email protected].

place at the solid-electrode-electrolyte interface. For example, there appears to be a wide gap between sophisticated catalysts which are now being synthesized for solution phase reactions9 and the difficulties encountered in the rational design of heterogeneous electrocatalysts. The reason for this may be in part the need to understand the critical factors controlling the rates of charge transport and ion transfer processes especially for reactions involving the transfer of multiple electrons,10 so that voltammetric studies on catalysts attached to electrodes may be very beneficial. The inherent potential complexity of reactions of solid materials attached to electrode surfaces can be readily understood by noting that processes such as (i) electron transfer at the electrode|solid interface,11 (ii) electron conduction by “hopping” or other types of mechanisms, (iii) ion-exchange at the solid|solution interface, (iv) ion transport, (v) exchange of

10.1021/jp9925942 CCC: $19.00 © 2000 American Chemical Society Published on Web 02/11/2000

1978 J. Phys. Chem. B, Vol. 104, No. 9, 2000

Bond et al. TABLE 1: Data Obtained from Cyclic Voltammograms (Scan Rate 0.1 V s-1) for the Oxidation of [{M(bipy)2}{M′(bipy)2}(µ-L)](PF6)n with M, M′ ) Ru or Os Dissolved in Dimethylformamide (0.1 M NBu4BF4) at a Platinum Disk Electrode (Taken from Ref 12)

RuRu OsRu OsOs a

Figure 1. Schematic representation of the molecular structure of [{M(bipy)2}{M′(bipy)2}(µ-L)]n+ with M ) Ru and M′ ) Ru.

neutral molecules between solid and solution phase, (vi) nucleation and relaxation phenomena, and (vii) chemical reaction steps coupled to the redox step can be present in any given system. Thus the range of possible effects originating from the structure of the solid is enormous and a more detailed knowledge of how the different processes are coupled is crucial in the understanding of solid-state redox chemistry. In the present study, dinuclear ruthenium and osmium metal complexes of the type [{M(bipy)2}{M′(bipy)2}(µ-L)](PF6)n with L ) 1,4-dihydroxy-2,5-bis(pyrazol-1′-yl)benzene (see Figure 1) are shown to undergo unusually fast reversible solid-state redox processes. The redox processes considered here are localized on the dioxolene bridge and correspond to a reversible hydroquinone, semiquinone, and quinone interconversions.12 Although oxidation state dependent structural changes cannot be ruled out, the dioxolene ligand will be denoted “L” independent of the oxidation state for reasons of simplicity. The crucial factors giving rise to these rapid solid-state processes, such as the structure of the solid and the type and concentration of the electrolyte anion, are investigated, and the main differences between solid and solution phase studies are identified. For the first time, in situ Raman spectroscopy is used in conjunction with the solid-state voltammetry of a microcrystalline material attached to the electrode surface to interrogate the solid-state reactions that occur at the electrode|solid|solution (electrolyte) interfaces. Experimental Section Reagents. The dioxolene-bridged dinuclear metal complexes [{M(bipy)2}{M′(bipy)2}(µ-L)](PF6)n with M/M′/n ) Ru/Ru/2 (RuRu), Ru/Os/3 (OsRu), and Os/Os/3 (OsOs) were prepared as described recently.12 For electrochemcial experiments, filtered and demineralized water from an Elgastat system (Elga, High Wycombe, U.K.) of resistivity of not less than 18 MΩ cm was used. Electrolyte salts used for the preparation of aqueous solutions were NaClO4, KCl, KSCN, KNO3, and KPF6 (all Aldrich, analytical, or the purest available quality). Solutions were purged with argon (Pureshield, BOC Gases) for at least 10 min prior to experiments. All experiments were conducted at 20 ( 2 °C. Electrode Preparation. For solid-state voltammetric measurements7 the solid sample (ca. 1 mg) was spread on a filter paper (Whatman 1) and the working electrode, a 4.9 mm diameter basal plane pyrolytic graphite disk (Pyrocarbon, Le Carbone Ltd., U.K.) mounted in a Teflon holder, was gently rubbed over the material causing some of it to adhere to the

process I: E1/2f,I/V vs SCE

process II: E1/2f,II/V vs SCE

∆E1/2f/V

E1/2f,I+II a/V vs SCE

0.13 0.19 0.23

0.54 0.44 0.42

0.41 0.25 0.19

0.335 0.315 0.325

E1/2f,I+II denotes the potential halfway between E1/2f,I and E1/2f,II.

electrode surface as a random array of particles. After use, the electrode surface was renewed by removal of the top graphite layers by repeated polishing on a fine carborundum paper. Instrumentation. Voltammetric experiments were undertaken with an Autolab PGSTAT 20 system (Eco Chemie, The Netherlands) in a conventional three-electrode electrochemical cell. The counter electrode was a gold or platinum wire and the reference a saturated calomel electrode (SCE, Radiometer Copenhagen). Scanning electron microscopy (SEM) was performed using a JEOL JSM-5200 system. For Raman studies, a Renishaw 1000 Raman microscope with a SpectraPhysics model 127 HeNe producing 633 nm red light at 35 mW was used. An integration time of 60 s was used to achieve a good signal-to-noise ratio. The in situ Raman electrochemical cell was made out of Teflon and allowed the working electrode to be placed in ca. 1 mm distance from the CaF2 window. Measurements were taken with a ×50 objective lens having a 8 mm focal length which produces a spot size of ca. 3 µm diameter. Results and Discussion Voltammetry of [{M(bipy)2}{M′(bipy)2}(µ-L)](PF6)n (n ) 2, 3) Attached to Graphite Electrodes and Immersed in Aqueous 0.1 M KPF6. Metal complexes of the type [{M(bipy)2}{M′(bipy)2}(µ-L)](PF6)2 with M, M′ ) Ru or Os and L ) the 1,4-dihydroxy-2,5-bis(pyrazol-1-yl)benzene dianion are known to undergo an extensive series of redox processes which have been characterized12 by electrochemical and spectroelectrochemical techniques (UV/vis and resonance Raman spectroscopy) in organic solvent solution-phase experiments. Four one-electron oxidation processes have been detected. In DMF or acetonitrile solution, the metal oxidation process (Ru(III/II) and Os(III/II)) has been observed at potentials positive of +1.0 V vs SCE and the bipyridyl ligand reduction at potentials negative of -1.0 V vs SCE. However, the two reversible bridging dioxolene ligand based processes, which produce the redox chemistry related to a hydroquinone system (see Figure 1), occur at potentials between 0.0 and 1.0 V vs SCE and are therefore within the potential range conveniently accessible with graphite electrodes in contact with aqueous electrolyte media. The oxidation of the bridging dioxolene ligand has been shown to occur in dimethylformamide solution in two one-electron steps12 (see Table 1) with the formation of an intermediate which may be described as a semiquinone (eqs 1 and 2).

[{M(bipy)2}{M′(bipy)2}(µ-L)]2+(solution) a [{M(bipy)2}{M′(bipy)2}(µ-L)]3+(solution) + e- (1) [{M(bipy)2}{M′(bipy)2}(µ-L)]3+(solution) a [{M(bipy)2}{M′(bipy)2}(µ-L)]4+(solution) + e- (2)

Fast Electron and Anion Transport Processes

J. Phys. Chem. B, Vol. 104, No. 9, 2000 1979

The semiquinone-type intermediate has been shown to be stabilized with respect to disproportionation by the presence of the metal fragments. The Os(bipy)22+ fragment has been found to exert a destabilizing effect compared to the Ru(bipy)22+ fragment. This causes the half-wave potentials (E1/2 values) for the processes described in eqs 1 and 2 to exhibit a 0.41 V potential gap compared to a potential gap of 0.19 V in the presence of two Os(bipy)22+ fragments. In contrast, the midpoint potential halfway between the two one-electron processes, E1/2f,I+II, is nearly independent of the metal fragment (see Table 1), which suggests that it is only the stability of the semiquinone that is affected by electronic effects.12 The solid-state voltammetric study of the metal complexes [{Ru(bipy)2}2(µ-L)](PF6)2 (RuRu), [{Os(bipy)2}2(µ-L)](PF6)3 (OsOs), and [{Ru(bipy)2}{Os(bipy)2}(µ-L)](PF6)3 (OsRu) focuses on the oxidation of the bridging dioxolene ligand when the electrode-solid interface is in contact with aqueous electrolyte, a medium in which both the oxidized and reduced forms of the compounds are essentially insoluble. Furthermore, provided the pH value remains close to neutral, as in this work, hydrolysis of metal-ligand bonds induced by acid is essentially absent. In Figure 2a, cyclic voltammograms obtained for the oxidation of [{Ru(bipy)2}2(µ-L)](PF6)2 mechanically attached to a basal plane pyrolytic graphite electrode and placed in aqueous 0.1 M KPF6 are shown. Two extremely well-defined redox processes with half-wave potentials at E1/2f,I ) 0.09 V vs SCE and E1/2f,II ) 0.38 V vs SCE are observed (see Table 2). These processes may be written as in eqs 3 and 4 where the ion transport reactions accompanying the charge transport process are omitted for simplicity.

[{M(bipy)2}{M′(bipy)2}(µ-L)]2+(solid) a [{M(bipy)2}{M′(bipy)2}(µ-L)]3+(solid) + e- (3) [{M(bipy)2}{M′(bipy)2}(µ-L)]3+(solid) a [{M(bipy)2}{M′(bipy)2}(µ-L)]4+(solid) + e- (4) The individual peak potentials measured as the average of the oxidation and reduction peak potentials are changing with scan rate, but as expected, the formal half-wave potentials, E1/2f, measured as the average of the oxidation and reduction peak potentials remain essentially constant. A plot of the peak potential versus the logarithm of the scan rate reveals a shift of 38 ( 5 mV per decade change in scan rate (scan rate range 10-200 mV s-1) for all four voltammetric responses. The peak currents increase almost linearly with the scan rate, which in conjunction with results from Raman spectroscopy (vide infra) implies that extensive electrolysis of a thin particulate deposit of compound occurs at the electrode surface. Further evidence for this interpretation of an overall two-electron electrolysis constructed from two closely spaced one-electron processes comes from the symmetric shape of the voltammetric responses and from the calculated integrated charge which over both oxidation processes shown in Figure 2a is 0.8 mC (corresponding to 6 µg of [{Ru(bipy)2}2(µ-L)](PF6)2) and independent of the scan rate over the range of 1 mV s-1 to 0.1 V s-1. The voltammetric responses obtained for the redox processes of solid [{Os(bipy)2}{Ru(bipy)2}(µ-L)](PF6)3 (Figure 2b) and [{Os(bipy)2}2(µ-L)](PF6)3 (Figure 2c) mechanically attached to basal plane pyrolytic graphite electrodes reveal the existence of analogous processes. For all three compounds, two chemically reversible processes occur within the potential range of interest and the individual formal half wave potentials, E1/2f, are

Figure 2. Cyclic voltammograms (scan rates 100, 50, 20, 10 mV s-1) for the oxidation of solid (a) [{Ru(bipy)2}2}(µ-L)](PF6)2, (b) [{Os(bipy)2}{Ru(bipy)2}(µ-L)](PF6)3, and (c) [{Os(bipy)2}2}(µ-L)](PF6)3 attached to basal plane pyrolytic graphite electrodes and then immersed in an aqueous 0.1 M KPF6 electrolyte. Cyclic voltammograms (scan rates 10 mV s-1) with Gaussian curves fitted for the oxidation of solid (d) [{Ru(bipy)2}2}(µ-L)](PF6)2, (e) [{Os(bipy)2}{Ru(bipy)2}(µ-L)](PF6)3, and (f) [{Os(bipy)2}2}(µ-L)](PF6)3.

1980 J. Phys. Chem. B, Vol. 104, No. 9, 2000

Bond et al.

TABLE 2: Data Obtained from Representative Cyclic Voltammetric Experiments (Scan Rate 0.01 V s-1) for the Redox Conversion of Solid [{M(bipy)2}{M′(bipy)2}(µ-L)](PF6)n with M, M′ ) Ru or Os Mechanically Attached to Basal Plane Pyrolytic Graphite Electrodes and Immersed in Aqueous 0.1 M KPF6a Epanod/V vs SCE

Epcath/V vs SCE

E1/2f b/V vs SCE

∆hhE1/2c/V

E1/2f,I+II d/V vs SCE

Ipanod/µA

Ipcath/µA

0.19 0.19

14 24

15 23

0.235

0.23 0.23

10 11

12 10

0.150

0.15 0.15

15 20

17 19

0.055

RuRu process I process II

0.17 0.44

0.0 0.31

0.09 0.38

process I process II

0.07 0.37

-0.06 0.25

-0.01 0.31

process I process II

0.05 0.18

-0.07 0.05

-0.01 0.12

OsRu

OsOs

a In order to analyze overlapping processes, Gaussian-shaped processes were fitted to the experimental data. b Formal half-wave potential calculated as the midpoint potential E1/2f ) 1/2 (Epanod + Epcath). c Peak width at half-height. d E1/2f,I+II denotes the potential halfway between E1/2f,I and E1/2f,II.

characteristically dependent on the metal fragments (see Table 2). In the solid state, the stabilizing nature of the ruthenium fragment on the semiquinone type radical intermediate can be clearly confirmed with ∆E1/2 ) 0.29 V for RuRu and ∆E1/2 ) 0.13 V for OsOs measured with a scan rate of 10 mV s-1. However, the value of the potential halfway between these two reversible half-wave potentials changes from 0.24 V (RuRu) to 0.15 V (RuOs) and finally to 0.06 V vs SCE (OsOs). Therefore, in contrast to the value of 0.32 V vs SCE (see Table 1) obtained in dimethylformamide, the metal fragment causes an additional shift in potential in the solid state reaction which is absent for the solution-phase redox process. This additional potential shift can arise from metal fragment dependent Gibbs free energy terms associated with the electroinsertion and expulsion of electrolyte anions between the aqueous and the solid phase. That is, the interaction between the anion and the metal complex, which is insignificant in the solution phase, becomes measurable in the solid state. The effect of changing the electroinsertion-undergoing anion is described in more detail below. It may be noted that, although some similarities of reversible potential data obtained in the organic solution phase and in the solid phase can be seen, the direct comparison is unwarranted due to the probable presence of significantly different junction potential effects. To minimize problems associated with calculation of peak heights and peak potentials from overlapping processes, the data reported in Table 2 are calculated on the basis of fitting Gaussian shaped redox waves to simulate the voltammograms observed at a scan rate of 10 mV s-1 (see Figure 2d-f). This procedure is justified by the thin layer nature of voltammetric wave shapes.13,14 The fact that the first redox wave for the RuRu system (process I) is smaller than the second redox wave (process II) is a complication, mirrored also in the data for OsOs, which is not explained by the exhaustively electrolyzed “thinlayer” model for each process. This effect diminishes at slower scan rates (Figure 2d). A possibly related effect has been discussed on the basis of interfacial charge transport effects for the solid-state redox conversion of chromium metal complexes.5 A further important parameter listed in Table 2 is the peak width at half-height, ∆hhEp, which is approximately constant for oxidation and reduction as well as for the first and the second redox waves. In a “thin-layer” model, the peak width at halfheight is a measure of interaction in the solid14 and should be 0.09 V for the ideal case of no interaction between the redox centers. If this thin-layer model is valid, then in all cases a repulsive interaction is attributed to the origin of broadening of the wave shape. There appears to be no change in the magnitude

of the interaction effect induced by incorporation of the first anion into the solid on the second redox process. Apparently, the initial redox state of the solid used for voltammetric analysis is not a crucial factor in the voltammetry. Thus, the mixed-metal material [{Os(bipy)2}{Ru(bipy)2}(µ-L)](PF6)3, which has been synthesized in the semiquinone rather than in the reduced state, also undergoes two reversible oneelectron redox processes (Figure 2b). The equilibrium potential (zero current potential) of the freshly prepared basal plane pyrolytic graphite electrode modified with the mixed-metal solid and immersed in aqueous 0.1 M KPF6 was found to be 0.073 V vs SCE. The fact that this value is intermediate to the formal half-wave potentials for processes I and II for OsRu (see Table 2) is consistent with the presence of the semiquinone. Commencing cyclic voltammetric experiments at the equilibrium potential of 0.073 V vs SCE does not lead to any significant changes in the voltammetric behavior. Remarkable stability of the electrochemical response for the oxidation of RuRu in aqueous 0.1 M KPF6 was shown via extensive potential cycling experiments. For example, continuous cycling with a scan rate of 0.1 V s-1 over a potential range from 0 to 0.9 V vs SCE for 100 cycles only resulted in a loss of peak current of e1%. This value is exceptionally low for experiments of this kind and may be interpreted in terms of (i) a high chemical inertness of the materials (in 0.1 KPF6) in all three redox states, (ii) very low water solubility at all oxidation levels, and (iii) the absence of redistribution processes which may give rise to complex features in solid-state voltammetric studies.15 Effect of KPF6 Electrolyte Concentration on the Voltammetry of [{Ru(bipy)2}2(µ-L)](PF6)2 Attached to Graphite Electrodes. Changing the electrolyte concentration in the aqueous phase may characteristically affect the voltammetric response obtained for solid microcrystalline materials adhering to the electrode surface and allows further insights into the details of the complex electrochemical solid-state process. In principle, there are two limiting cases in which the half-wave potential, E1/2f, is independent of the electrolyte concentration (“thin-film model”) or in which a characteristic Nernstian shift of 59 mV in E1/2f (at 25 °C) per decade change in concentration of the anion for a one-electron process is observed (“thick-film model”). The nuances of these and other related theoretical models have been discussed.16,17 The effect of changing the concentration of the aqueous electrolyte on the voltammetric responses for the oxidation and rereduction of solid [{Ru(bipy)2}2(µ-L)](PF6)2 attached to a basal plane pyrolytic graphite electrode and immersed in KPF6 is shown in Figure 3. The peak

Fast Electron and Anion Transport Processes

J. Phys. Chem. B, Vol. 104, No. 9, 2000 1981

Figure 4. Schematic representation of a solid particle of [{M(bipy)2}{M′(bipy)2}(µ-L)](PF6)n adhered to the surface of a graphite electrode.

Figure 3. (a) Cyclic voltammograms (scan rate 10 mV s-1) for the oxidation of solid [{Ru(bipy)2}2}(µ-L)](PF6)2 attached to basal plane pyrolytic graphite electrodes and then immersed in aqueous 0.1, 0.02, and 0.005 M KPF6 electrolyte. (b) Plot of the peak potentials (Epanod and Epcath) versus the logarithm of the concentration of KPF6. The dashed lines show the shift in half-wave potential.

potentials are almost linearly dependent on the logarithm of the concentration of KPF6 (see Figure 3b) although the anodic and cathodic peaks are affected differently with the reduction peak potentials being less dependent on concentration than the oxidation potentials. The shift in half wave potentials observed for process I is 41 mV and for process II 52 mV per decade change in electrolyte concentration, which is less than expected on the basis of eqs 5 and 6 and use of the appropriate Nernst equation (59 mV at 25 °C).

[{M(bipy)2}{M′(bipy)2}(µ-L)](PF6)2(solid) + PF6-(solution) a [{M(bipy)2}{M′(bipy)2}(µ-L)](PF6)3(solid) + e- (5) [{M(bipy)2}{M′(bipy)2}(µ-L)](PF6)3(solid) + PF6-(solution) a [{M(bipy)2}{M′(bipy)2}(µ-L)](PF6)4(solid) + e- (6) Clearly, neither the “thin-layer model” nor the “Nernst model” is completely valid. A more realistic schematic representation of a particle of redox-active material adhering to an electrode surface is shown in Figure 4. The partitioning of the anions (and possibly also of the cations) between the solid and solution phase causes a charged layer to be formed at the solid|solution interface analogous to a “Donnan” potential.17b,c A parallel may be drawn between this case of small microcrystalline particles and that at an organic solvent|aqueous electrolyte interface18 which lead to sub-Nernstian behavior attributed to ion partitioning. However, conditions for the system with particles adhering to the electrode surface clearly are more complex due to the presence of the three-phase boundary.

The origin of the increase in the gap between the peak potentials for oxidation and reduction as the supporting electrolyte concentration decreases (Figure 3b) is unknown but probably is the result of a kinetic effect associated with the electron or ion transport process. With a thin-layer model, a gap between the anodic and cathodic peak potential19 may arise from either slow electron transfer across the electrode|solid interface or a slow rate of ion transport across the solution|solid interface. Effect of the Nature of the Electrolyte on the Voltammetry of [{M(bpy)2}{M′(bpy)2}(L)](PF6)n Attached to Graphite Electrodes. The extremely fast oxidation and electroinsertion of PF6- anions into the structure of [{M(bipy)2}{M′(bipy)2}(µ-L)](PF6)n type materials is unexpected on the basis of previous solid-state voltammetric studies of metal complexes.7,8 Commonly, voltammetric responses obtained in the 0.1 V s-1 scan rate range have lead to only minor interconversion of the attached and reduced forms of solid. In other systems nucleationgrowth type mechanisms have been found to be operative causing a substantial overpotential or an inert zone.20 However, polymeric inorganic structures, such as Prussian Blue,6 have been shown to allow rapid ion insertion/expulsion and substantial redox state conversion within such short period of time. These latter classes of materials may be considered as “electrochemically open” in the sense that minimal energy is required to achieve the structural change necessary for the accommodation of electrons and ions in the solid and consequently in these circumstances both mass and electrical transport can be fast. Synthesis of [{M(bipy)2}{M′(bipy)2}(µ-L)](PF6)n compounds does not yield well-defined crystalline compounds, and inclusion of solvent and even salts (KNO3) has been detected.12 SEM images of RuRu (Figure 5) reveal that these solids readily disintegrates into small submicrometer-sized particles, which implies that the forces holding the solid together are rather weak. However, despite this structural feature, the solid does not undergo dissolution in aqueous media due to the large size and the hydrophobic nature of the cation. Shifts of more than 0.25 V may be detected in the formal potential (changing from PF6- to NO3-) for processes involving the transfer of the anion from aqueous to organic phases.21 However, in the case of [{M(bipy)2}{M′(bipy)2}(µ-L)]2+ voltammetric data summarized in Table 3 show that only small shifts toward more positive potential are observed in the formal potential, E1/2f, when changing the electrolyte from relatively hydrophobic PF6- to the relatively hydrophilic NO3-. This result suggests that the difference in Gibbs free energy for transfer of the anion from the aqueous to the solid phase is small, as expected when the interaction between the electroinserted anion and the host lattice is similar to that between the anion and water. The separation of oxidation and reduction peak potentials

1982 J. Phys. Chem. B, Vol. 104, No. 9, 2000

Bond et al.

Figure 6. Raman spectra obtained for [{Os(bipy)2}2(µ-L)](PF6)3 (OsOs) attached to a basal plane pyrolytic graphite and then immersed in 0.1 M KPF6 shown after the potential has been stepped to the indicated value and held for 10 s.

Figure 5. SEM images of solid [{Ru(bipy)2}2}(µ-L)](PF6)2 attached to sticky tape shown in (a) high and (b) lower magnification.

TABLE 3: Data Obtained from Representative Cyclic Voltammetric Experiments (Scan Rate 0.1 V s-1) for the Redox Conversion of Solid [{M(bipy)2}{M′(bipy)2}(µ-L)](PF6)n with M, M′ ) Ru or Os Mechanically Attached to Basal Plane Pyrolytic Graphite Electrodes and Immersed in Aqueous 0.1 M KPF6, NaClO4, KSCN, or KNO3 Electrolyte Solution RuRu

OsRu

f/V

-

PF6

ClO4SCNNO3-

process I process II process I process II process I process II process I process II

f/V

E1/2 vs SCE

∆Ep/V

E1/2 vs SCE

∆Ep/V

0.09 0.38 0.12 0.39 0.14 0.36 0.19 0.39

0.23 0.22 0.20 0.21 0.21 0.21 0.07 0.10

-0.01 0.31 0.03 0.34 0.04 0.35 0.03 0.37

0.26 0.22 0.15 0.17 0.10 0.06 0.07 0.08

also becomes very small when NO3- is the anion, implying the kinetics of transport of this anion across the interface is extremely fast. The above-mentioned ready formation of the KNO3-containing mixed salt12 also may be important in achieving a very fast rate of charge neutralization. If K+ cations are present in the reduced form of the compound, loss of K+ as well as the uptake of NO3- from the solution could lead to the charge neutralization required during the course of oxidation, and this may be especially significant in the presence of low concentrations of aqueous electrolyte. Trends observed in voltammetric experiments with OsOs when the electrolyte is varied are consistent with the results described above for the RuRu and OsRu cases. However, the partial overlap of the two oxidation processes together with the less stable nature of the voltammetric responses increases the uncertainty in the analysis of the data.

In Situ Controlled-Potential Spectroelectrochemical Raman Studies of the Voltammetry of [{M(bipy)2}{M′(bipy)2}(µ-L)](PF6)n Attached to Graphite Electrodes. Spectroelectrochemical methods are essential for the elucidation of the mechanism and identification of products of electrochemical processes. In solid-state voltammetric experiments, in which only microgram quantities of materials are electrolyzed and converted to materials with a different composition and redox state, the range of in situ spectroelectrochemical approaches available to aid the understanding of the processes is limited. In situ Raman spectroscopy has not, to date, been applied to this area of electrochemistry even though its superior flexibility offers considerable advantages over other optical methods. The HeNe laser employed in these measurements (wavelength ) 633 nm) may be focused onto small spots of 3 µm diameter on the electrode surface and also easily penetrates aqueous electrolytes and conventional glass or quartz windows used in electrochemical cells. In Figure 6, a typical sequence of Raman spectra obtained during the course of oxidation or rereduction of [{Os(bipy)2}2(µ-L)](PF6)3 (OsOs) mechanically attached to a basal plane pyrolytic graphite and immersed in aqueous 0.1 M KPF6 electrolyte is shown. The intense signals at 1580 and 1331 cm-1, also detected in the absence of attached compound, are attributed to the basal plane pyrolytic graphite background. The change in spectra observed in this sequence of experiments clearly demonstrates both the reversibility and significant extent of the solid-state conversion well beyond the monolayer level. Importantly, the small laser spot size of 3 µm and experiments at several different points on the electrode surface allow measurements without extensive spatial averaging and preclude the possibility of unreacted islands in the bulk material remaining during the course of these experiments. Since the Raman measurements carried out in this investigation refer to pyrolyitic graphite as opposed to metal electrodes, no signal enhancement due to SERS is anticipated. The 633 nm excitation precludes resonance enhancement of Ru or Os to bipy MLCT transitions which occur at shorter wavelengths.12 However, the laser wavelength is postresonant with a band previously assigned as a hydroquinone-bipy interligand transition12 at approximately 580 nm (see Figure 6 and Table 4), and it is changes associated predominantly with this transition

Fast Electron and Anion Transport Processes

J. Phys. Chem. B, Vol. 104, No. 9, 2000 1983

TABLE 4: Raman Spectral Data (in cm-1) Obtained during the Course of in Situ Experiments in Which Solid [{M(bipy)2}{M′(bipy)2}(µ-L)](PF6)n with M, M′ ) Ru or Os Attached to a Basal Plane Pyrolytic Graphite Electrode and Immersed in Aqueous 0.1 M KPF6 Electrolyte Is Electrolyzed by Setting the Potential (E) of the Working Electrode at the Indicated Value RuRu OsRu OsOs

reduced form

semiquinone form

oxidized form

E ) -0.5 V vs SCE: 1554, 1486, 1273, 1155, 1022, 663, 223 E ) -0.5 V vs SCE: 1480, 1265, 1166, 1020, 830, 750, 670, 200 E ) -0.5 V vs SCE: 1473, 1259, 1166, 1045, 1014, 778, 745, 673, 400, 372, 200

E ) 0.25 V vs SCE: 1486, 1273, 1155, 1036, 663, 223 E ) 0.25 V vs SCE: 1480, 1265, 1166, 1020, 830, 750, 670, 200 E ) 0.06 V vs SCE: 1473, 1259, 1241 1166, 1045, 1014, 778, 745, 727, 673, 400, 372, 200

E ) 0.75 V vs SCE: 1495, 1036, 736, 663, 654, 550, 372, 223 E ) 0.75 V vs SCE: 1500, 830, 750, 600, 200 E ) 0.6 V vs SCE: 1491, 1241, 1045, 1014, 745, 727, 673, 641, 609, 595, 200

that are observed. This feature is manifested by the loss of bipybased bands at 1464 and 1158 cm-1 after complete oxidation of the bridge. A mode observed at ca. 660 cm-1, which is associated with the bridging hydroquinone type ligand coupled to a Os-O mode, decreases as the first oxidation step occurs and is eliminated on complete oxidation of the bridge. Importantly, this band at 660 cm-1 is replaced by a band at 716 cm-1 associated with generation of the Os-OdC bridge-coupled mode present in the quinone form of the ligand and also detected in previous solution phase studies carried out on these complexes.12 This observation is therefore important in verifying both the bridge-based nature of the solid-state electrochemical processes under study and the fact that the sequence of redox events follows those observed in solution. This same pattern of Raman spectra changes is observed for OsRu and RuRu (see Table 4). The metal-independent nature of these spectral changes further confirms the bridge- rather than metal-based nature of these electrochemical steps. Raman experiments conducted in the presence of aqueous 0.1 M NaClO4 electrolyte gave spectra generally similar to those obtained with 0.1 M KPF6. However, due to the higher solubility of compounds formed in the presence of ClO4-, which is enhanced at elevated temperatures induced by heating from the laser, the Raman signal intensities for this system diminished with time. Conclusions The electrochemical oxidation of [{M(bipy)2}{M′(bipy)2}(µ-L)](PF6)n type compounds (M, M′ ) Ru or Os) is characterized by two chemically reversible one-electron-transfer steps and rapid bulk conversion of solid on the electrode surface which is accompanied by uptake and expulsion of ions from the aqueous electrolyte. While the two oxidation processes occur at potentials that are related to those observed in solution-phase voltammetric experiments, the dependence of the formal halfwave potentials on the type and concentration of the electrolyte is characteristic of the solid-state processes. The conclusion that [{M(bipy)2}{M′(bipy)2}(µ-L)](PF6)n solids are “electrochemically open” has been reached on the basis of (i) the high currents observed from essentially bulk conversion of the solid on the voltammetric time scale and the absence of smaller “surface” signals,22 (ii) the absence of nucleation phenomena associated with the formation of new phases, and (iii) the continuous change in redox level which allows the equilibrium potential of the interface to be easily established. The form of structural features responsible for the fast electron and ion transport and the behavior approaching that expected for “thin-layer” type voltammetry appear to be a molecular structure that readily accommodates reversible electron and ion transfer, large molecular ions of low symmetry which aid Coulombic cation-anion-electron interaction and ion exchange across the solid-aqueous solution boundaries,

structural packing which favors salt or solvent inclusion, and low solubility of both the oxidized and reduced forms of the redox couple. Acknowledgment. Christiaan H. Goeting and Dr. John S. Foord are gratefully acknowledged for their assistence with the SEM measurements, and Dr. Colin D. Bain is acknowledged for assistance with Raman experiments. T.E.K. acknowledges generous support from Johnson Matthey, and F.M. thanks the Royal Society for the award of a University Research Fellowship and New College, Oxford, U.K., for a Stipendiary Lectureship. References and Notes (1) See for example: Fiedler, D. A. J. Solid State Electrochem. 1998, 2, 315. (2) See for example: Solid State Electrochemistry; Bruce, P. G., Ed.; Cambridge University Press: Cambridge, U.K., 1995. (3) Scholz, F.; Meyer, B. Electroanal. Chem. 1998, 20, 1. (4) Scholz F.; Lange, B. Chem. Soc. ReV., 1994, 341. (5) Bond, A. M.; Colton, R.; Daniels, F.; Fernando, D. R.; Marken, F.; Nagaosa, Y.; Van Steveninck, R. F. M.; Walter, J. N. J. Am. Chem. Soc. 1993, 115, 9556. (6) Dostal, A.; Meyer, B.; Scholz, F.; Schro¨der, U.; Bond, A. M.; Marken, F.; Shaw, S. J. J. Phys. Chem. 1995, 99, 2096. (7) Bond, A. M.; Marken, F. J. Electroanal. Chem. 1994, 372, 125. (8) Bond, A. M.; Colton, R.; Marken, F.; Walter, J. N. Organometallics 1994, 13, 5122. (9) See for example: Aqueous-Phase Organometallic Catalysis; Cornils, B., Hermann, W. A., Eds.; Wiley-VCH: New York, 1998. (10) See for example: Steiger, B.; Shi, C.; Anson, F. Inorg. Chem. 1993, 32, 2107. (11) Lange, B.; Lovric, M.; Scholz, F. J. Electroanal. Chem. 1996, 418, 21. (12) Keyes, T. E.; Forster, R. J.; Jayaweera, P. M.; Coates, C. G.; McGarvey, J. J.; Vos, J. G. Inorg. Chem. 1998, 37, 5925. (13) Bara, A. J.; Faulkner, L. R. Electrochemical Methods; Wiley: New York, 1980. (14) Laviron, E. Electroanal. Chem. 1982, 12, 53. (15) Bond, A. M.; Fletcher, S.; Symons, P. G. Analyst 1998, 123, 1891. (16) See for example: Armstrong, F. A.; Heering, H. A., Hirst, J. Chem. Soc. ReV. 1997, 26, 169. (17) (a) Naegeli, R.; Redepenning, J.; Anson, F. C. J. Phys. Chem. 1986, 90, 6227. (b) Redepenning, J.; Tunison, H. M.; Moy, J. J. Phys. Chem. 1994, 98, 2426. (c) Redepenning, J.; Miller, B. R.; Burnham, S. Anal. Chem. 1994, 66, 1560. (18) (a) Shi, C.; Anson, F. C. J. Phys. Chem. B 1999, 103, 6283. (b) Kakiuchi, T. Anal. Chem. 1996, 68, 3658. (19) See for example: Electrochemical Methods; Bard, A. J., Faulkner, L. R., Eds.; Wiley: New York, 1980; p 412. (20) Bond, A. M.; Fletcher, S.; Marken, F.; Shaw, S. J.; Symons, P. G. J. Chem. Soc., Faraday Trans. 1996, 92, 3925. (21) Marken, F.; Compton, R. G.; Goeting, C. H.; Foord, J. S.; Bull, S. D.; Davies, S. G. Electroanalysis 1998, 1, 821. (22) Bond, A. M.; Cooper, J. B.; Marken, F.; Way, D. M. J. Electroanal. Chem. 1995, 396, 407.