Solvent dynamic effects in electron transfer: electrochemical versus

---

J . Phys. Chem. 1990, 94, 2949-2954

2949

Solvent Dynamical Effects in Electron Transfer: Electrochemical versus Self-Exchange Kinetics of Tris(hexafluoroacetylacetonato)ruthenium(I 1111I ) and Comparison with Other Probe Reactants M . J. Weaver,* D. K. Phelps, R. M. Nielson, M. N. Golovin, and G . E. McManis Department of Chemistry, Purdue University, West Lafayette. Indiana 47907 (Received: July 28, 1989)

Rate constants for electrochemical exchange, k&, for Ru(hfac):/- (hfac = hexafluoroacetylacetonate) at annealed gold in six solvents are reported and compared with solvent-dependent rate constants, k:,, for Ru(hfac):/- self-exchange and with corresponding data for metallocene and related organometallic redox couples. The Ru(hfac):/- couple was selected to explore the possible differences in the solvent-dependent dynamics resulting from the presence of such large aliphatic, rather than small aromatic, ligands. The solvents examined-acetonitrile, acetone, nitromethane, benzonitrile, nitrobenzene, and propylene carbonate-display a range of longitudinal relaxation times, T~-',enabling the sensitivity of the kinetics to T~ to be assessed and intercompared with those for other reactions. In contrast to the organometallic systems which display facile 7,-dependent electrochemical kinetics, characteristic of adiabatic pathways, the k:, values for Ru(hfac):/- are markedly (ca. 102-fold) smaller and insensitive to the solvent dynamics, correlating instead with the anticipated solvent-dependent energetics. This latter behavior is consistent with markedly nonadiabatic pathways, where the preexponential factor is solvent independent, being controlled by the (subunity) electron-tunneling probability rather than by nuclear dynamics. The self-exchange kinetics of Ru(hfac),O/-, although more facile than for electrochemical exchange, also display a similar insensitivity to the solvent dynamics. Further evidence that the sluggish electrochemical exchange kinetics of Ru(hfac):/- primarily reflect nonadiabatic pathways, rather than the effects of inner-shell reorganization, was obtained by intercomparisons of k:, and k!, for Ru(hfac):/with corresponding rate data for redox couples known to exhibit substantial inner-shell barriers and from the differing solvent dependencies of k:, for these systems. The virtues of examining such combined rate dependencies upon solvent, reactant structure, and reaction environment for distinguishing between electronic and nuclear reorganization factors in electron transfer are noted.

A number of recent experimental studies from this laboratory'.* and have focused on the manner and extent to which solvent dynamics ("solvent friction") can influence the barriercrossing frequency for electron-transfer reactions. Our initial examinations of solvent-dependent electrochemical kinetics of metallocenium-metallocene and other redox couples' have been supplemented more recently by detailed studies of a number of related homogeneous self-exchange reactions2 The latter class of systems offer several advantages for this purpose; in particular, the availability of optical electron-transfer energies for related binuclear systems provides direct experimental estimates of the required solvent-dependent barrier heights, AG*.6 A question of central significance concerns the degree to which the influence of solvent dynamics upon the electron-transfer rates can be sensitive to the reactant structure and environment as a consequence of variations in the donor-acceptor orbital interaction~.'.~When this "orbital overlap", as gauged by the electronic ( I ) (a) Weaver, M. J.; Gennett, T. Chem. Phys. Lett. 1 9 8 5 , l l 3 , 2 1 3 . (b) Gennett, T.; Milner. D. F.;Weaver, M. J. J . Phys. Chem. 1985.89, 2787. (c) McManis, G. E.; Golovin, M. N.; Weaver, M. J. J . Phys. Chem. 1986, 90, 6563. (d) Nielson, R . M.; Weaver, M. J. J . Electroanal, Chem. 1989, 260, 15. (e) Nielson, R. M.; Weaver, M. J . Organometallics 1989, 8, 1636. (2) (a) Nielson, R. M.; McManis, G . E.; Golovin, M. N.; Weaver, M. J. J . Phys. Chem. 1988, 92, 3441. (b) Nielson, R. M.; McManis, G. E.; Safford, L. K.; Weaver, M. J. J . Phys. Chem. 1989, 93, 2152. (c) Nielson, R. M.; McManis, G . E.; Weaver, M. J. J . Phys. Chem. 1989, 93, 4703. (d) McManis, G. E.; Nielson, R. M.; Gochev, A,; Weaver, M. J . J . Am. Chem. Soc. 1989, 1 1 1 , 5533. (3) Other electrochemical studies include: (a) Kapturkiewicz, A,; Behr, B. J . Elecfroanal. Chem. 1984, 179, 187. (b) Kapturkiewicz, A.; Opallo, M. J . Electroanal. Chem. 1985, 185, 15. (c) Kapturkiewicz, A. Electrochim. Acta 1985, 30, 1301. (d) Opallo, M. J . Chem. Soc., Faraday Trans. I 1986,82, 339. (4) Other homogeneous-phase thermal electron-transfer studies include: (a) Harrer, W.; Grampp, G.; Jaenicke, W. Chem. Phys. Left. 1984, 112, 263. (b) Harrer, W.; Grampp, G.; Jaenicke, W. J . Elecrroanal. Chem. 1986, 209, 223. (c) Kapturkiewicz, A,; Jaenicke, W. J . Chem. Sor., Faraday Trans. I 1987, 83, 16 I . (5) Photoinduced homogeneous-phase studies include: (a) Simon, J. D.; Su,S.-G. J . Phys. Chem. 1988, 92, 2395. (b) Kahlow, M. A,; Kang, T. J.; Barbara, P. F. J . Phys. Chem. 1987, 91, 6452. (c) Kang, T. J.; Kahlow, M. A.; Giser, D.; Swallen, S.;Nagarajan, V.; Jarzeba, W.; Barbara, P. F. J . Phys. Chem. 1988, 92, 6800. (6) McManis, G . E.; Gochev, A.; Nielson, R . M.; Weaver, M. J . J . Phys. Chem. 1989, 93, 7733.

0022-3654/90/2094-2949$02.50/0

matrix coupling element, HI2,is relatively small (say 50.1 kcal mol-'), the barrier-crossing frequency is predicted to be controlled primarily instead by the "electron-hopping probability" so that the reaction rates will be insensitive to, and ultimately independent of, the solvent dynamics.' Indeed, the achievement of conditions approaching reaction adiabaticity, whereupon the full influence of solvent dynamics is felt, is expected to require high-friction media and/or relatively large degrees of orbital overlap, say H I 2 k 0.5 kcal In addition, the role of solvent dynamics is predicted to be muted for reactions featuring moderate or large inner-shell (Le., intramolecular) barriers9 One therefore might expect that the importance, or even the qualitative presence, of solvent dynamical effects would be sensitively dependent upon the system structure. We have recently demonstrated such a sensitivity of the ratesolvent friction dependence,2das well as the rates themselves,I0 to reactant electronic structure for a series of metallocene selfexchange reactions featuring alterations in the ligand substituents and the central metal, yet where the inner-shell barriers are uniformly small. Analysis of the rate-friction dependencies enabled estimates of HI, to be obtained, which are in accord with corresponding theoretical predictions.2d On the other hand, the electrochemical exchange kinetics for some of these couples are insensitive to reactant electronic structure and uniformly strongly dependent upon the solvent dynamics.lbSc This differing behavior of such related electrochemical and homogeneous-phase exchange processes indicates that the former proceed by inherently more adiabatic pathways, in harmony with the relatively strong electronic coupling predicted recently for electron transfer at metal surfaces." Given that similar behavior has also been noted for (7) (a) McManis, G . E.; Mishra, A. K.; Weaver, M. J. J . Chem. Phys. 1987,86, 5550. (b) Gochev, A,; McManis, G . E.; Weaver, M. J. J . Chem. Phys. 1989, 91, 906. (8) (a) Beratan, D.N.; Onuchic, J . N. J . Chem. Phys. 1988.89, 6195. (b) Onuchic, J. N.; Beratan, D. N . J . Chem. Phys. 1988, 92, 4818. (9) (a) Sumi, H.; Marcus, R. A. J . Chem. Phys. 1986, 84, 4894. (b) Nadler, W.; Marcus, R. A. J . Chem. Phys. 1987, 86, 3906. (IO) Nielson, R. H.; Golovin, M. N.; McManis, G. E.; Weaver, M. J . J . Am. Chem. Soc. 1988, 110, 1745. (1 I ) (a) Morgan, J. D.; Wolynes, P. G. J . Phys. Chem. 1987, 91, 874. (b) Zusman, L. D.Chem. Phys. 1987, 112, 53.

0 I990 American Chemical Society

2950

The Journal of Physical Chemistry, Vol. 94, No. 7, 1990

several other electrochemical exchange reaction^,^^%*^ it is reasonable to inquire what factors might conspire to yield sufficiently weak surface-reactant orbital overlap so to engender nonadiabatic electrochemical pathways. Most electrochemical exchange processes examined so far in this context involve electron transfer to and from aromatic (or organometallic) reactants that might be expected to yield substantial orbital overlap with the electrode, even for outer-sphere processes where a layer of solvent molecules separates the reactant from the metal surface. However, reactants where the metal redox center is surrounded by relatively large aliphatic ligands are anticipated to yield weaker orbital overlap. The range of known candidate systems is not large, especially when the requirements of solubility in a range of solvents and a monocharged/uncharged redox couple (so to minimize work terms) are taken into account. One suitable couple, however, is tris(hexafluoroacety1acetonato)ruthenium(III/II) [abbreviated here as Ru(hfac),O/-], in that the -CF3 substituents should aid the "insulation" of the metal center. A key virtue of this system is that the self-exchange kinetics have received careful examination in a number of solv e n t ~ ; ' ~analysis ,'~ of these results indicates a virtual absence of solvent dynamical effects (vide infra). Reported here are rate constants for R~(hfac)~O/electrochemical exchange at an annealed gold surface in six solvents, chosen so to yield variations in dynamical as well as energetic solvation factors and to engender comparison with the observed kinetics of Ru(hfac),O/- self-exchange. The results indicate that this reaction proceeds via sluggish nonadiabatic pathways in the electrochemical environment, as deduced in part from the insensitivity of the kinetics to solvent dynamics as well as from the rates themselves. This deduction is supported by comparisons with solvent-dependent kinetic data for some related electrochemical exchange and homogeneous self-exchange reactions. Experimental Section

Tris( hexafluoroacetylacetonato)ruthenium( 111) (Strem Chemicals) was purified by vacuum sublimation. Tetraethylammonium perchlorate (TEAP, G.F. Smith Chemicals) was recrystallized from hot ethanol. Tetrabutylammonium hexafluorophosphate (TBAH) was prepared by metathesis from the bromide salt (Aldrich) and ammonium hexafluorophosphate (Ozark-Mahoning) in acetone. Purification of acetonitrile, acetone (Burdick & Jackson), nitromethane (Mallinckrodt), nitrobenzene, and benzonitrile (Fluka) followed standard procedure^;'^ propylene carbonate (Burdick & Jackson) was used without further purification. The rate constants for electrochemical exchange, k&, were obtained by using either phase-selective ac voltammetry or cyclic voltammetry. The former procedure was mostly as described in refs 1 b and IC. This utilized a PAR 173/179 potentiostat, a PAR 175 potential programmer, a PAR 5204 lock-in amplifier, a Hewlett-Packard 3314A function generator, and a Fluka 1900A frequency counter. Positive-feedback IR compensation was employed. Measurements of the phase angle, 4, as a function of ac frequency, W, over the range 50-1000 Hz when plotted as cot 4 versus ail2yielded kzx values.tb.c The required diffusion coefficients, D,were obtained from the cathodic dc polarographic limiting currents, using a dropping mercury electrode. Measurements were made in an argon atmosphere away from direct exposure t o light.

Weaver et al. TABLE I: Observed Rate Constants for Electrochemical Exchange, k,,, and Related Parameters for Ru(hfac)JO/-at Gold Electrodes in Various Solvents at 23 OC electroEf,6 105D,C k,,,d rL-l,r no. solvent lyte" V vs SCE cm2 s-I cm s-I I O i 2 s-I 1 acetonitrile TEAP 0.71 1.3 0.019 4 TBAH 0.009 2 acetone TEAP 0.84 1.4 0.026 3.5 3 nitromethane TEAP 0.66 0.8 0.035/# 4.5 4 benzonitrile TEAP 0.74 0.35 0.089 0.2 TBAH 0.109 5 nitrobenzene TEAP 0.61 0.25 0.069 0.2 6 propylene TEAP 0.2 0.48 0.02/ (0.4) carbonate Electrolyte concentration was 0.1 M. TEAP = tetraethylammonium perchlorate; TBAH = tetrabutylammonium hexafluorophosphate. * Formal potential for R~(hfac)~O/in medium specified, determined by means of ac voltammetry and/or cyclic voltammetry. Diffusion coefficients of Ru(hfac)3, determined from the dc polarographic limiting current for R ~ ( h f a c )reduction. ~ dRate constant for Ru(hfac)t/- electrochemical exchange, determined by ac voltammetry and/or cyclic voltammetry as noted (see text). 'Inverse solvent longitudinal relaxation time at 25 O C , taken from ref 2d (see ref I C and Table I of ref 2c for literature sources and calculational details). Value for propylene carbonate given in parentheses since this solvent displays additional relaxation(s) at higher frequencies.2a-c /From cyclic voltammetry. 6 From ac voltammetry.

XY recorder or a Nicolet Explorer I digital oscilloscope. Although 1R compensation was employed for cyclic as well as ac voltammetric measurements, where both techniques could be used, the former tended to yield significantly (ca. 2-fold) smaller k:, values. Most probably, this disparity is due to the inferior degree of IR compensation that can be achieved under cyclic voltammetric conditions.'6 More consistent results, however, could be obtained by using a modified procedure whereby subtracted from the sweeprate-dependent cathodic-anodic peak separations were the values obtained under identical conditions for the ferroceniumferrocene (Cp2Fe+io)couple, rather than the "ideal" reversible value, 57 mV. Since corresponding ac voltammetric measurements yield k:, values for Cp2Fe+loin most solvents that are close to the upper limit even of this technique (ca. 1-5 cm s-'), the difference between the cyclic voltammetric peak separations for Cp,Fe+/O and 57 mV is probably due largely to uncompensated solution resistance. The gold working electrodes were ca. I-mm-diameter beads, joined to gold wires sealed in glass. They were pretreated immediately prior to use by heating to just below the melting point (where "surface melting" visibly occcurs) in a propane-oxygen flame. This pretreatment was found to yield unusually reproducible and stable electrochemical kinetic behavior (cf. refs Id and le). Roughly comparable k;, values for Ru(hfac),O/-, within ca. 2-3-fold, were also obtained by using other pretreatment procedures, such as mechanically polishing gold disks and cycling the potential in the chosen solvent prior to adding the reactant. The gold annealing procedure was, however, preferred here after extensive evaluation of these alternatives in view of the excellent reproducibility of the resulting kinetics. The R~(hfac)~O/formal potentials are too positive to enable mercury electrodes to be employed for the k:, measurements. All electrode potentials were measured and are quoted versus an aqueous saturated calomel electrode (SCE), and all kinetic measurements were made at room temperature (23 f 1 "C).

For k:, values below ca. 0.03-0.05 cm s-' (depending on the diffusion coefficient), the ac responses were insufficient to yield reliable results. T h e r a t e constants were then obtained instead by using cathodic-.anodic cyclic voltammetry at sweep rates from 0.2 to 2 V s-', employing the analysis procedure of N i c h o l ~ o n . ~ ~ Results and Discussion Table 1 contains a summary of kEx values for Ru(hfac),O/- in The voltammograms were displayed on a Hewlett-Packard 7045 six solvents: acetonitrile, acetone, nitromethane, benzonitrile, nitrobenzene, and propylene carbonate. Two main factors de(12) Chan, M A ;Wahl, A. C. J . Phys. Chem. 1982, 86, 126. termined (and limited) the choice of media employed. The positive (13) Doine, H.; Swaddle, T. W. Inorg. Chem. 1988, 27, 665. formal potentials, El, for Ru(hfac),O/- (20.5 V versus SCE, Table (14) (a) Perrin, D. D.; Armarcgo, W. L. F.; Perrin, D. R. Purification of Laboratory Chemicals, 2nd ed.; Pergamon: New York, 1980. (b) Riddick. I ) eliminated from consideration a number of relatively oxidizable

J . A.; Bunger, W. B.; Sakano, T. K. Organic Soluents-Physical Properties and Methods of Purification, 4th ed.; Wiley: New York. 1986. (IS) Nicholson. R. S. Anal. Chem. 1965, 37, 1351

(16) Milner, D. F.: Weaver, M. J. Anal. Chim. Acta 1987. 198, 245.

Solvent Dynamical Effects in Electron Transfer

The Journal of Physical Chemistry, Vol. 94, No. 7, 1990 2951

solvents, such as amides. A primary concern is to select solvents having a wide range of dynamical characteristics (cf. refs 1 and 2). While restricted somewhat by the above limitation, the extent of solvent "friction" as gauged by the longitudinal relaxation time, 7L, is substantially larger in benzonit,rile and nitrobenzene than in the "low-friction" media acetonitrile, acetone, and nitromethane, also included in Table Another consideration is that largely the same media have been employed in two of our earlier, related, studies of solvent-dependent electrochemical kinetics,ldscthereby facilitating the data intercomparison which is pursued below. The k:, values listed in Table I were typically reproducible to within 25% or so, for replicate determinations using at least three different gold electrodes. Although increasing the concentration of TEAP supporting electrolyte from 0.1 to 0.3 M yielded little rate change (525%), significant effects upon k:, were seen in some cases upon substituting tetraethylammonium (TEA) by tetrabutylammonium (TBA) cations (Table I). Such rate dependencies upon the supporting electrolyte cation are commonly observed for anionic redox couples." While the presence of such effects admittedly clouds somewhat the quantitative data interpretation, the insensitivity of k:, to the ionic strength suggests that the magnitude of the electrostatic double-layer corrections, or at least their solvent dependence, is sufficiently small to be neglected for the present semiquantitative purposes (cf. refs Id and le). Inspection of the rate data in Table I reveals two features of particular interest. First, the k:, values themselves fall in the range ca. 0.01-0.1 cm s-l, which are much smaller than those (20.5 cm s-l) commonly observed for metallocene and related redox couples that feature small inner-shell (i.e., bond distortional) barriers, AG*k.l One initially plausible explanation for this lower reactivity is that the Ru(hfac)30/- couple involves significant inner-shell activation, such that AG*i, 2-3 kcal mol-', associated with stretching the ruthenium-oxygen bonds upon reduction of Ru(II1) to Ru(I1). Some support for this notion is derived from the observation that AG*i, for R U ( O H & ~ + / ~electrochemical + exchange is estimated by crystallographic data to be ca. 2 kcal Unfortunately, insufficient bond distance data are available for Ru(hfac),O/- or related complexes to yield reliable AG*i, estimates. Evidence presented below, however, indicates that inner-shell activation is unlikely to be the major origin of the small k:, values for Ru(hfac),O/-. The other significant feature in Table I concerns the solvent dependence of k:,. Most outer-sphere redox couples for which reliable solvent-dependent electrochemical kinetic studies have been performed so far yield k:, values that correlate at least qualitatively with the solvent dynamics as monitored by TL-'.'J However, no such correlation is seen in Table I; indeed, the two solvents having the smallest TL-' (i.e., the greatest "solvent friction"), benzonitrile and nitrobenzene, also yield the largest k:, values for Ru(hfac),O/-. Analysis of Rate-Solvent Dependencies. Such k:,-solvent dependencies are at least roughly consistent with the anticipated variations in the outer-shell (Le., solvent reorganization) energy, AG*,. To explore this point further, as well as to analyze solvent-dependent kinetic data from a more general standpoint, it is convenient to express rate constants for either electrochemical exchange, k:,, or homogeneous self-exchange, k:,, as20,21

-

k,, = K p ~ , Iexp[-(AG*i, ~,

+ AG*,)/RT]

is the electronic transmission coefficient. The last two terms, K,', describe the frequency at which the transition state is approached and the probability that electron transfer will occur once the barrier top is reached, respectively. For adiabatic I ; since it is often expected that u, 7L-l (vide pathways, K,' infra),'J the solvent dependence of k,, can then be controlled primarily by the variations in the unimolecular preexponential factor ~ , p ,rather than merely the solvent energetics as embodied in AG*,. For nonadiabatic pathways (K,'