3680
J . Phys. Chem. 1985, 89, 3680-3684
Luminescence of Mixed-Ligand Polypyridine-Ruthenium( II)Complexes in the Temperature Range 84-250 K. Interltgand Interactions and Viscosity Effects on Radiationless Processes F. BarigelIetti,la P. Belser,lb A. von Zelewsky,lb A. Juris,la,cand V. Balzani*la,C Istituto di Fotochimica e Radiazioni d’Alta Energia del CNR, Bologna, Italy, Istituto Chimico “G. Ciamician” dell’universita’, Bologna, Italy, and Institute of Inorganic Chemistry, University of Fribourg, Fribourg, Switzerland (Received: November 29, 1984; In Final Form: March 5, 1985)
The luminescence behavior (emission spectra and emission lifetimes) of Ru(bpy),(LL)3-,2+ mixed-ligand complexes ( n = 1 or 2; bpy = 2,2’-bipyridine; LL = 2,2’-biquinoline (biq), 2,2’-biisoquinoline (i-biq), or a 2,2’-biquinoline derivative (DMCH) and of the parent Ru(bpy)?+ complex has been studied between 84 and 250 K, a temperature range which includes the rigid-fluid transition temperature of the solvent (4:5 (v/v) propionitrilebutyronitrile). As the temperature increases, the emission maxima of the luminescence spectra undergo a red shift in the rigid-fluid transition region, and then a blue shift at higher temperature. The luminescence lifetime decreases with increasing temperature in all the temperature ranges examined, with a strong discontinuity in the glass-fluid transition region; the luminescence intensity exhibits a qualitatively similar behavior. The In I/T vs. 1/T plots are fitted over all the temperature range by an equation which contains a temperature-independent term k,, an Arrhenius term A exp(-AEIRT), and a term B/(1 + exp[C(l/T- l/TB)]Jwhich takes care empirically of the behavior in the glass-fluid transition region. The results obtained indicate that the radiationless deactivation processes are markedly dependent on viscosity. This suggests that very-large-amplitude (low-frequency) vibrational modes, which are most likely Ru-N skeletal motions, are involved in the radiationless deactivationprocesses as promoting and accepting modes. A comparison of the results obtained for the Ru(bpy)2(LL)2+and R u ( ~ ~ ~ ) ( L Lcomplexes ) ~ * + shows that the interaction between biq or DMCH ligands leads to measurable changes in the kinetic parameters and also to changes in the energy levels. The problem of localized vs. delocalized excitation in the Ru(bpy),(LL),_;+ complexes is discussed.
Introduction
The study of the photophysical properties of transition-metal complexes is currently a topic of great interestZ because of theoretical reason^^-^ and potential applications in the field of energy Ru(bpy)3z+(bpy = 2,2’-bipyridine) has played a key in the development of this research area, but presently much attention is focussed on other Ru-polypyridine complexes, with particular emphasis on those containing different types of ligands.’OJ’ Temperature-dependence studies of the luminescence behavior can yield important information concerning energy, electronic nature, and deactivation rate of the luminescent and reactive excited ~ t a t e s . ~ - ~ ~ When l ~ ~ ’ such ~ - ’ ~studies are performed (1) (a) Istituto FRAE-CNR. (b) University of Fribourg. (c) University of Bologna. (2) Demas, J. N. J . Chem. Educ. 1983, 60, 803. (3) Crosby, G . A. Acc. Chem. Res. 1975, 8, 231. (4) Kemp, T. J. Prog. React. Kinet. 1980, 10, 301. (5) DeArmond, M. K.; Carlin, C. M. Coord. Chem. Rev. 1981,36, 325. (6) Graetzel, M., Ed. “Energy Resources through Photochemistry and Catalysis”; Academic Press: New York, 1983. (7) Balzani, V.; Bolletta, F.; Ciano, M.; Maestri, M. J . Chem. Educ. 1983, 60, 447 and references therein. (8) Kalyanasundaram, K. Coord. Chem. Reu. 1982,46, 159. (9) Watts, R. J. J . Chem. Educ. 1983, 60, 834. (10) The literature in this field is very rich. Most of the previous papers are quoted in the following references: (a) Allen, G. H.; White, R. P.; Rillema, D. P.; Meyer, T. J. J. Am. Chem. SOC.1984,106,2613. (b) Cherry, W. R.; Henderson Jr., L. J. Inorg. Chem. 1984, 23, 983. (c) Morris, D. E.; Ohsawa, Y.;Segers, D. P.; DeArmond, M. K.; Hanck, K.W. Inorg. Chem. 1984,23, 3010. (d) Juris, A,; Barigelletti, F.; Balzani, V.; Belser, P.; Von Zelewsky, A. Inorg. Chem. 1985, 24, 202. (1 1) Balzani, V.; Juris, A,; Barigelletti, F.; Belser, P.; Von Zelewsky, A. Riken Q. 1984, 78,78. This paper contains a table where the photophysical and electrochemical properties of some 50 Ru-polypyridine complexes are listed. (12) Wrighton, M.; Morse, D. L. J . Am. Chem. SOC.1974, 96, 998. (13) Van Houten, J.; Watts, R. J. J . Am. Chem. SOC.1976, 98, 4853. (14) Watts, R. J.; Harrington, J. S.; Van Houten, J. J . Am. Chem. SOC. 1977, 99, 2179. (15) Giordano, P. J.; Frederick, S . M.; Wrighton, M. S . ; Morse, D. L. J . Am. Chem. SOC.1978, 100, 2257. (16) Watts, R. J.; Missimer, D. J . Am. Chem. SOC.1978, 100, 5350. (17) Caspar, J. V.; Meyer, T. J. J . Am. Chem. SOC.1983, 105, 5583. (18) Barigelletti, F.; Juris, A,; Balzani, V.; Belser, P.; Von Zelewsky, A. Inorg. Chem. 1983, 22, 3335, and unpublished results.
0022-3654/85/2089-3680$01.50/0
over a sufficiently large temperature range, a rigid-fluid transformation of the matrix takes place which may cause a change in the spectral distribution of l u m i n e s c e n ~ e ~(luminescence ~~l~-~~ rigidochromismi2) and discontinuity in the 1 / vs. ~ 1/T In this paper we show that mixed-ligand polypyridine-ruthenium(I1) complexes exhibit a marked change in their luminescence properties in the rigid-fluid transition region and that the interpretation of such a behavior may yield useful pieces of information to elucidate some excited-state properties. Experimental Section
PF6- salts of the R U ( ~ ~ ~ ) , ( L L ) complexes ~ - , , ~ + were prepared and purified as described e 1 s e ~ h e r e . l ~In the general formula given above, n is 1 or 2, bpy stands for 2,2’-bipyridine, and LL is 2,2’-biisoquinoline (i-biq), 2,2’-biquinoline (biq), or a 2,2’-biquinoline derivative (DMCH). The structural formulae and abbreviations of the various ligands are shown in Figure 1. All the experiments were carried out in a mixture of freshly distilled propionitrile and butyronitrile (4:5 v/v). A diluted solution (10-5-10-4 M) of each complex was sealed under vacuum in an 1-cm quartz cell after repeated freeze-pump-thaw cycles. The cell was then placed inside a Thor C 600 nitrogen flow cryostat, equipped with a Thor 3030 temperature controller and indicator. The absolute error on the temperature is estimated to be f 2 K. The emission spectra were obtained with a Perkin-Elmer MPF-44B spectrofluorimeter equipped with a Hamamatsu R 928 phototube. Emission lifetimes were measured by a modified Applied Photophysics single photon counting apparatus equipped with a thyratron gated source lamp. The decay was monitored at the maximum of the emission band. Data treatment was carried out with a PDP/ 11 Digital microcomputer. Standard iterative nonlinear programs were employed to analyze the decay curves.2o The quality of the fit was assessed by the xz values close to unit and the residuals regularly distributed along the time axis. The relative emission intensities at the various temperatures were obtained from the area of the emission bands. Because of the change in the transparency of the matrix as well as in other parameters in the (19) Belser, P.; Von Zelewsky, A. Helo. Chim. Acta 1980, 63, 1675. (20) Bevington, P. R. ‘Data Reduction and Error Analysis for Physical Sciences”; McGraw-Hill: New York, 1969.
0 1985 American Chemical Society
The Journal of Physical Chemistry, Vol. 89, No. 17. 1985 3681
Polypyridine-Ruthenium(I1) Complexes
DMCH
biq
Figure 1. Structural formulae and abbreviations of the ligands.
600
800
X.nm
'O0
Figure 4. Emission spectra of Ru(bpy),2' (a), R U ( ~ ~ ~ ) , ( D M C (b), H)~' and Ru(bpy)(DMCH)?' (c). Full line, at 100 K; dashed line, at 150 K. The emission intensities have been arbitrary scaled for the sake of easier comparison.
I
F
I A
'O0
X,nm
700
Figure 2. Emission spectra of Ru(bpy),*' (a), Ru(bpy)z(i-biq)2' (b), Ru(bpy)(i-biq)?' (c), and Ru(i-biq)?' (d). Full line, at 100 K dashed line, at 150 K. The emission intensities have been arbitrary scaled for the sake of easier comparison.
0 0
a
0 00 0
L , , , . , 4
8
1*
i000/~.~-
Figure 5. Change in the wavelength of the emission maximum with temperature. A: Ru(bpy)2(DMCH)2', B: Ru(bpy)32t. The other mixed-ligand complexes examined exhibit a qualitatively similar behavior.
800
700
800
I
X ,nm Figure 3. Emission spectra of Ru(bpy)32' (a), Ru(bpy)2(biq)zt(b), and Ru(bpy)(biq),2' (c). Full line, at 100 K; dashed line, at 150 K. The
emission intensities have been arbitrary scaled for the sake of easier comparison. rigid-fluid transition, no quantitative analysis was performed on these intensity data. Results Figures 2-4 show the emission spectra of the mixed ligand R~(bpy),(i-biq)~-,'+, Ru(bp~),(biq)~-,*+, and Ru(bpy),-
(DMCH)>? complexes below ( T = 100 K) and above ( T = 150 K) the glass-fluid transition region. For comparison purposes, the spectra of Ru(bpy)y and Ru(i-biq)?' are also shown. Except for Ru(i-biq)j2', on increasing temperature the emission maxima of the luminescence spectra undergo a red-shift in the rigid-fluid transition region and a blue shift at higher temperature. Typical examples of this behavior are shown in Figure 5. In each case the emission spectrum was independent of the excitation wavelength. The temperature dependence of the emission lifetime in the range 84-250 K for Ru(bpy),2t10d and for the mixed-ligand bpy-biq and bpy-DMCH complexes is shown in Figure 6. The mixed-ligand complexes containing bpy and i-biq behave in the same way, as shown previously.". For all the complexes examined, the change in the luminescence intensity with temperature is roughly parallel to the lifetime changes, as shown in Figure 6 for Ru(bpy)32+. In all cases the emission was found to decay according to a single exponential expression. The change in the experimental emissiQn rate k = 1 / with ~ increasing temperature can be accounted for by the equation k = ko + B/(1 + e x p [ C ( l / T - l/TB)]) + A exp(-AE/RT) (1)
3682 The Journal of Physical Chemistry, Vol. 89, No. 17, 1985
Barigelletti et al.
excited states involving the ligand that is easier to reduce, which is the LL ligand when LL = biq or DMCH and bpy when LL = i-biq.1“,21,22In other words, in the mixed-ligand complexes one type of ligand is “luminactive” and the other type plays the role of “spectator”. It follows that: (a) for all the complexes discussed in this paper (Table I) emission originates from 3MLCT excited 1’ states with the only exception of Ru(i-biq),,’; (b) for the Ru(bpy),(i-biq),_:+ series (n = 1-3), the MLCT emission only n involves the bpy ligands; (c) for the Ru(bpy),(biq),->+ and RuW C (~P~)~(DMCH)~-:+ series (n = 1 or 2), the MLCT emission only involves the biq or DMCH ligands. The R ~ ( b i q ) , ~and + Ru(DMCH)?’ complexes are not included in the following discussion 1 because of complicating features deriving from the presence of low-energy metal-centered excited states.’* A problem which is still the object of debate is whether the emitting MLCT excited states of R ~ ( b p y ) , ~(and, + by analogy, \ of R ~ ( b i q ) , ~and + Ru(DMCH),,+) are best described as multichelate ring delocalized with molecular D3 symmetry or single e ‘ 1 -e-* chelate ring localized (spatially isolated) with C, ~ y m m e t r y .In ~ the last few years, evidence has accumulated in favor of a localized 4 8 IOOO/T,K-? de~cription.~~-~* This, of course, does not exclude a small amount of interligand intera~tion,~ but this aspect of the problem has not Figure 6. Temperature dependence of emission lifetime for R ~ ( b p y ) , ~ ’ yet been elucidated. (01, Ru(bpy)dbiq)2’ (A),Ru(bpy)(biq),2’ (4,R ~ ( ~ P Y ) ~ ( D M C H ) ~ ’ Rigid Matrix Effects. The temperature change in emission ( O ) ,and R U ( ~ ~ ~ ) ( D M C (D). H ) ~ The ~ ’ temperature dependence of the lifetimes in the 84-250 K temperature range was found to obey relative emission intensity for Ru(bpy)?’ (*) is also shown. The intensity eq 1. The term ko represents the rate limit value at 84 K, and behavior for the other complexes is qualitatively similar and is not shown for the sake of clearness. it receives contributions from radiative and radiationless comp o n e n t ~ . The ~ ~ Arrhenius-type term is present both below and above the temperature range of the rigid-fluid transition: this TABLE I: Kinetic Parameters for Excited-State Decay Obtained shows that it correspoods to an intrinsic molecular process, subfrom the Fitting of Eq 1 to the Experimental Results‘ stantially unaffected by solvent viscosity. In agreement with complexb kO; s-I B, s-l A, s-I cm-I previous discussion^,^ the activation energy AE may be assigned Ru(bpy)(i-biq)?’ 2.9 X lo5 2.3 X lo5 7.8 X lo5 110 as the energy separation between thermally equilibrated emitting 65 Ru(bpy)z(i-biq)2t 3.1 X lo5 2.8 X los 4.8 X lo5 ~ 1/T in the temperature levels. The stepwise behavior of 1 / vs. 1.6 X lo6 9.0 X l o 5 8.0 X lo7 770 Ru(bp~Mbiq)~’ range of the rigid-fluid transition was satisfactorily accounted Ru(bpy)(biq)?’ 6.4 X l o 5 2.0 X l o 5 3.0 X lo6 300 for by the term 1
+ -
bQ%
R t ~ ( b p y ) ~ ( D M c H ) ~1.2 + X R u ( b p y ) ( D M C H ) F 6.8 X Ru(bPY)32+ 3.2 x 1.9 x Ru(i-biq),*’ e
lo6 lo5 105 104
6.4 3.3 2.1 8.9
lo5 2.6 X 10’ lo5 1.2 X lo6 x 105 5.6 x 105 x 103 X X
630 190 90
“The parameters C and TB were found to vary between 1900 and 4000 and 120 and 125 K, respectively. bThe emitting excited states are M L C T in all cases, except for Ru(i-biq)32’. CThe experimental error on the emission lifetime is 58%. dEstimated error 520%. e Ligand-centered emission.
where ko is a temperature-independent term (at 84 K), the second term takes care empirically of the effect of the rigid-fluid transition, and the third term is the usual Arrhenius expression which contains a frequency factor and an activation barrier. The parameters ko, B, C, TB, A , and AE have been obtained from the nonlinear iterative fitting to the experimental points and are reported in Table I. In the same table, the results obtained for Ru(bpy)$+ and Ru(i-biq),,+ have also been reported for the sake of comparison.
Discussion It has long been known that the luminescence emission of R ~ ( b p y ) ~takes ~ ’ place from a cluster of closely lying, formally triplet metal-to-ligand charge-transfer (MLCT) excited state^.^ The luminescence emission of Ru(biq),,+ and RU(DMCH),~+is also due to formally triplet MLCT excited states,21,22 whereas the luminescence of Ru(i-biq),,+ is a ligand-centered (LC) phosp h o r e ~ c e n c e . ~As ~ far as the mixed-ligand complexes are concerned, in all cases emission only originates from “triplet” MLCT (21) Juris, A.; Balzani, V.;Belser, P.; Von Zelewsky, A. Hela Chim. Acta 1981, 64, 2175. (22) Juris, A.; Barigelletti, F.; Balzani, V.;Belser, P.; Von Zelewsky, A. Isr. J . Chem. 1982, 22, 87. (23) Belser, P.; Von Zelewsky, A,; Juris, A.; Barigelletti, F.; Tucci, A,; Balzani, V. Chem. Phys. Left. 1982, 89, 101.
This empirical term may be considered as an additional contribution, kd, to the decay of the luminescent excited state(s) which depends on temperature (Le., viscosity) so as to be equal to a congtant, B, for T m and equal to zero for T 0. According to eq 2, TB is the temperature at which ko’ = B/2 and C is a temRerature related to the viscosity effect to be discussed below. In principle, this deactivating term kd might receive contributions from radiative and radiationless components. We have found that the decrease in lifetime in the rigid-fluid transition region is accqmpanied by a similar decrease in the emission intensity (Figure 6). Although a quantitative comparison between the lifetime and intensity changes cannot be done because of experimental difficulties (see Experimental Section), it is apparent that the main (if not the only) contribution to ko’ comes from nonradiative deactivation processes. This means that for our complexes there is a radiationless deactivation path which is to some extent “frqzen” when the solvent matrix is rigid and which begins to be important only as the solvent matrix becomes fluid. Such a behavior can be accounted for by using the concept of intramolecular perturbation of molecular potentials discussed in detail by Dellinger and
-
-
(24) Carlin, C. M.; DeArmond, M. K. Chem. Phys. Left. 1982, 79, 297. (25) Braterman, P. S.;Harriman, A,; Heath, G.A,; Yellowlees, L.J. J . Chem. SOC.,Dalton Trans. 1983, 1801. (26) Bradley, P. G.; Kress, N.; Hornberger, B. A,; Dallinger, R. F.; Woodruff, W. H. J. Am. Chem. Soc. 1981, 103, 7441. (27) Forster, M.; Hester, R. E. Chem. Phys. Lett. 1981, 81, 42. (28) Kober, E. M.; Sullivan, B. P.; Meyer, T. J. Inorg. Chem. 1984, 23, 2098. (29) The lifetime vs. temperature changes at very low temperature for Ru(bpy)?’ are very c ~ m p l i c a t e d . ~ * ~ ~ (30) Ferguson, J.; Herren, F. Chem. Phys. 1983, 76, 45.
Polypyridine-Ruthenium(I1) Complexes
The Journal of Physical Chemistry, Vol. 89, No. 17, 1985 3683
Kasha31 and previously used to explain rigid matrix effects in the Ru(bpy)(i-biq)z2+-Ru(bpy)z(i-biq)2+ couple (Figure 2), the s p e c t r o ~ c o p y . ~ ~The * ~ ~viscosity ~ ~ * of the rigid matrix may be emission maxima coincide within the experimental error, sugthought to make steeper the vibrational potential well and/or gesting a weaker interligand interaction. smaller the equilibrium nuclear distances. Large-amplitude In the absence of interligand interactions one would also expect (low-frequency) vibrational modes are expected to be particularly the same ko, A , and AE values for complexes belonging to the sensitive to the presence of such a rigid matrix perturbation. It same R U ( ~ ~ ~ ) , ( L L ) ~family - ? + (n = 1 or 2, LL = biq, DMCH, follows that the Ru-N stretching and bending vibrations may be or i-biq). As one can see from the values collected in Table I, strongly affected by the rigid-fluid transition. These vibrations this is not the case for LL = biq or DMCH, which confirms the span all the nontotally symmetric irreducible representations of conclusion reported above concerning a noticeable interligand the C2, group to which the Ru-LL subunit belongs,Iw and thus interaction in these complexes. For the bpy ligands, the interaction they possess the right symmetry requisites to promote radiationless appears to be so weak as to be almost undetectable (within the deactivation of "forbidden" excited states.33 It follows that, when estimated experimental errors). However, some evidence for the rigid matrix "freezes" such Ru-N vibrations the complex may interligand interaction between the bpy ligands has been previously obtained from a comparison of the temperature dependence of exhibit, to some extent, a "vibrationally deficient" behavior in the sense described by Gardner and K a ~ h a with , ~ ~a consequent inintensity and lifetime.Iw A larger interligand interaction between crease in the luminescence intensity and lifetime. the biq and DMCH ligands is not unexpected. Interligand interaction may occur through space (Le., by direct orbital overlap) Another peculiar effect of the rigid matrix perturbation is or through bonds (Le., via orbital overlap mediated by the central apparent from the shift of the emission maximum with changing metal atom). Compared with bpy, the biq and DMCH ligands temperature (Figure 5 ) . The blue shift observed with increasing are expected to interact better both through space because they temperature in the region above the rigid-fluid transition may are more sterically crowding, and through bonds because their be associated with the involvement of higher energy emissive levels ?r* orbitals are closer in energy to the d ( r ) metal orbitals, giving (vide supra). In the rigid-fluid transition region, however, another rise to more covalent M-L bonds. process predominates which displaces the emission maximum in As pointed out by Meyer and co-workers,28 the problem of the opposite direction (rigidochromic effect).12 This process is localization vs. delocalization of excited states in complexes likely the relaxation of the above discussed rigid matrix perturcontaining more than one luminactive ligand may be discussed bation which causes a broadening of the potential well and/or an in terms of mixed-valence system^.^^,^^ According to this apincrease in the equilibrium nuclear distances in the excited state. proach, the key factors are the relative magnitudes of the deloAs one can see from Figures 2-4, (i) the red-shift is almost calization energy arising from electronic coupling between the negligible for Ru(i-biq)j2+, as expected for a ligand localized electron donor and acceptor ligands, and the vibrational trapping excited state which is not affected by perturbations on large-amenergy which receives contributions from solvent repolarization plitude (low-frequency) metal-ligand vibrational modes; (ii) it and from intramolecular normal modes for which there are is larger and constant when the ligand involved in the MLCT changes in equilibrium coordinates or frequencies. For the transition is bpy, in agreement with the smaller size of the bpy bpy/bpy- self-exchange in Ru"(bpy),+ ligand and with a localized excitation (vide infra); (iii) in the case of the R~(bpy),,(LL)~-,2+ complexes with LL = biq or DMCH, it is larger for n = 2 than for n = 1, Le., when the excitation has to be localized on a single ligand (vide infra). The kinetic pathe activation energy in CH3CN solution is estimated to be loo0 rameter B also shows analogous trends. cm-1.40941 The delocalization of the electron in the * R ~ ( b p y ) ~ , + It should be noted that on matrix melting solute-solvent inexcited state teractions (e.g., solvent repolarization, charge transfer, etc.) can also come into play, enhancing radiationless deactivations. Znterligand Interaction. Information on interligand interaction is an essentially analogous process which should thus exhibit a can be obtained from a comparison between the luminescence quite similar activation energy. On the other hand, comparison behavior of the R~(bpy),(biq)~+, R U ( ~ ~ ~ ) , ( D M C Hand ) ~ +Ru, among the emission spectra of Ru(bpy),(i-biq),-;+ (n = 1-3, (bpy)(i-biq)?+ complexes, where there is only one luminactive Figure 2) shows that the bpy-bpy interaction is small enough to ligand, and the R~(bpy)(biq)~,+, Ru(bpy)(DMCH)z2+,and Ruhave no apparent effect on the spectroscopic levels (within an (bp~),(i-biq)~+ complexes where two identical luminactive ligands experimental uncertainity of a few tens of cm-I). Thus, in the are present. In the absence of interligand interaction one would *R~(bpy),(i-biq)~-:+ complexes ( n = 2 or 3) the delocalization expect exactly the same behavior for each couple of complexes energy is much smaller than the vibrational trapping energy (class belonging to the same family (Le., for R U ( ~ ~ ~ ) , ( L and L ) ~Ru+ I1 compounds in Robin and Day's n ~ m e n c l a t u r e ~and, ~ ) on the ( ~ P Y ) ( L L ) ~ , +A) . very weak interaction may have consequences vibrational relaxation time scale, excitation is localized on a single on the deactivation rate constants, whereas a stronger interaction bpy ligand, in agreement with the previously available evidence can also affect the energy level^.^^^^^ From Figures 3 and 4 one of localized description for * R ~ ( b p y ) , ~ + However, . ~ ~ - ~ ~ for the can see that the emission maxima of Ru(bpy)(biq)?+ and RuR~(bpy),,(LL)~-n2+ complexes (n = 0 or 1, LL = biq or DMCH) ( ~ P ~ ) ( D M C H ) are , ~ +appreciably shifted compared with the emission maxima of R~(bpy)~(biq),+ and R U ( ~ ~ ~ ) , ( D M C H ) ~ +where the luminactive ligands are biq or DMCH, the situation may be quite different for two reasons. Firstly, the vibrational both in fluid and in rigid matrix, showing that the interligand trapping energy is expected to be smaller than that of bpy because interaction is strong enough to affect the spectroscopic levels. For of the larger size of the biq and DMCH ligands. Secondly, interligand interaction is strong enough to affect the spectroscopic levels to a noticeable extent (hundreds of cm-', see Figures 3 and (31) Dellinger, B.; Kasha, M. Chem. Phys. Lett. 1975, 36, 410; 1976, 38, 4 and Table I). It follows that vibrational trapping energy and 9. delocalization energy might have comparable magnitudes, so that, (32) Kasha, M.; Dellinger, B., Brown, C. In "2nd International Symposium on the vibrational relaxation time scale, the excitation might not on Analytical Applications of Bioluminescence and Chemiluminescence, San
-
Diego, CA, 1980"; De Luca, M. A,; Mc Elroy, W. D., Eds.; Academic Press: New York, 1981. (33) These vibrations may also act as accepting modes34for the same radiationless deactivation processes. (34) Caspar, J. V.; Meyer, T. J. Inorg. Chem. 1983, 22, 2444. (35) Gardner, P. J.; Kasha, M. J . Chem. Phys. 1969, 50, 1543. (36) Foerster, Th. In "Modern Quantum Chemistry"; Part 111; Sinanoglu, O., Ed.; Academic Press: New York, 1965. (37) Baggott, J. E.; Gregory, G. K.; Pilling, M. J.; Anderson, S.;Scddon, K. R.; Turp, J. E. J . Chem. SOC.,Faraday Trans. 2 1983, 79, 195.
(38) Brown, D. B., Ed."Mixed-ValenceCompounds";Reidel: Dordrecht, Holland, 1980. (39) Creutz, C. Prog. Inorg. Chem. 1983, 30, 1 . (40) Motten, A. G.; Hanck, K.; DeArmond, M. K. Chem. Phys. Lett. 1981, 79, 541. (41) Heath, G. A.; Yellowlees, L. J.; Braterman, P. S. Chem. Phys. Lett. 1982, 92, 646. (42) Robin, M. B.; Day, P. Adv. Inorg. Chem. Radiochem. 1967, IO, 241.
3684
J . Phys. Chem. 1985,89, 3684-3689
be fully localized on a single luminactive ligand.
blica Istruzione and by the Swiss National Science Foundation.
Acknowledb“ent. We thank Mr. L. Minghetti and Mr. L. Ventura for technical assistance. This work was supported by the Italian National Research Council and Minister0 della Pub-
Registry No. Ru(bpy)(i-biq)?+, 89340-69-2; Ru(bpy)z(i-biq)2+, 87464-48-0; Ru(bpy)z(biq)2+,75777-90-1; Ru(bpy)(biq)p, 75785-58-9; Ru(bpy)z(DMCH)2+, 75778-00-6; Ru(bpy)(DMCH),2+, 75778-02-8; Ru(bpy),2+, I5 158-62-0; Ru(i-biq)j2+, 82762-29-6.
Photoredox Reactions on Semiconductors at Open Clrcuit. Reduction of Fe3+ on WO, Electrodes and Particle Suspensions Jean Desilvestro* and Michael Neumann-Spallart* Institut de Chimie Physique, Ecole Polytechnique FZdZrale de Lausanne, CH- 101 5 Lausanne, Switzerland (Received: December 14, 1984; In Final Form: April 15, 1985)
The photoreactions at polycrystalline W 0 3 electrodes in Fe3+containing solutions are investigated. Photoreduction rates of Fe3+on unbiased W 0 3 are shown to be quantitatively related to current-potential curves in the dark and under illumination at the same electrode. One reaction product, Fez+, competes with H 2 0 for holes in the valence band of the semiconductor. From the electrochemical results, rates in illuminated W 0 3 particle suspensions can be estimated and compare well to values obtained by product analysis. The influence of Fe3+and Fe2+concentration, light intensity, and mass transfer is discussed.
Introduction
The idea of producing “useful” chemical substances by photochemical synthesis on semiconductor materials has shown to be fruitful for the principal understanding of semiconductor electrochemistry. Besides photoelectrochemical cells with solid electrodes, particle suspensions became interesting for the purpose of conversion and chemical storage of energy (for a review see ref 1). However, until very recently the analysis of observed photoproduction rates in particle suspensions has not met quantitatively the insights as yielded from electrochemical research on macroelectrodes. In 1967 Freund et aL2 suggested the investigation of photocatalytic and photosynthetic reactions on suspended semiconductor particles by studying separately cathodic and anodic processes on semiconductor electrodes. Since then astonishingly little work has been performed in this direction. Yoneyama et aL3 and Hada et aL4 found a fairly good correlation between product formation rates and local cell currents for unbiased illuminated TiOz and ZnO electrodes (reduction of methylene blue or Ag’, oxidation of CH30H, H 2 0 , or ZnO). In our work on TiOzSand WO; we have shown that rates of photoredox reactions like reduction of methylviologen (MV2+)and Ag+ can well be related to the internal currents in unbiased electrodes. These currents can be predicted from potentidynamic measurements on the same electrodes with the help of the microelectrode theory of local elements that has already been successfully applied for the understanding of redox catalysis by noble metals’ and corrosion phenomena.* The (1) Kalyanasundaram, K. In ‘Energy Resources through Photochemistry and Catalysis”, GrPtzel, M.,Ed.; Academic Press: New York, 1984. (2) (a) Morrison, S. R.; Freund, T. J. Chem. Phys. 1967,47, 1543. (b) Freund, T.; Gomes, W. P. Catal. Rev. 1969, 3, 1 . (3) Yoneyama, H.; Toyoguchi, Y.; Tamura, H. J . Phys. Chem. 1972, 76, 3460. (4) (a) Hada, H.; Tanemura, H.; Yonezawa, Y . Bull. Chem. SOC.Jpn. 1978, 51, 3154. (b) Hada, H.; Yonezawa, Y.; Ishino, M.; Tanemura, H. J. Chem. SOC.,Faraday Trans. 1 1982, 78, 2671. ( 5 ) Enea, 0.;Neumann-Spallart, M. J. Electrochem. Soc. 1984,131,2767. (6) Erbs, W.; Desilvestro, J.; Borgarello, E.;GrHtzel, M. J . Phys. Chem. 1984,88, 4001. (7) (a) Wagner, C.; Traud, W. 2.Z . Elekrrochem. 1938, 44, 391. (b) Miller, D. S.;McLendon, G . J . Am. Chem. SOC.1981, 103, 6791. (c) Sutcliffe, E.;Neumann-Spallart, M. Helv. Chim. Acta 1981, 64, 2148. (d) Albery, W. J.; Bartlett. P. N.; McMahon, A. J. In ’Photogeneration of Hydrogen”, Harriman, A., West, M. A., Eds.; Academic Press: London, 1982. (8) Evans, U. R. In ‘The Corrosion and Oxidation of Metals”: E.Arnold: London, 1960.
0022-3654/85/2089-3684$01 SO10
performance of a particle suspension of TiOz illuminated through a window of the geometry of a previously investigated electrode (same irradiance) could then be compared to rates on the electr~de.~ In this work we want to continue our studies on the dynamics of photoredox reactions on semiconductor electrodes at open circuit and particle suspensions with the intention of investigating in detail the influence of mass transfer and concentration of electroactive species. This influence will be shown to be exerted on the dark polarization characteristics and to determine the values of the overall production rates on unbiased semiconductor assemblies under illumination. By comparing results obtained on electrodes and particle suspensions we will try to show how the topological differences between the two systems influence their photoresponse. The system W03/Fe3+ was chosen because both reaction products (Fez+, 0,) of the photogenerated charge carriers can easily be followed analytically and the onset potential of Fe3+ reduction is positive enough to allow for significantly high photoproduction rates. This photoreaction has been described previously by Krasnovskii9 and by Mills.lo Bard” studied qualitatively the photoreduction of Fe3+on illuminated TiOz suspensions by in situ polarography. In this paper we will present an improved technique for measuring quantitatively the product formation rates at irradiated electrodes and particle suspensions. Experimental Section W 0 3 (Johnson & Matthey, 99.998%, 4.1 mZ/g), Fe(N03)3.
9 H 2 0 , FeS04-7Hz0,70% HC104 (Fluka, analytical grade), and K N 0 3 (Merck, analytical grade) were used as supplied. Deionized water was distilled twice. The preparation and characterization of polycrystalline W 0 3 electrodes have been reported previously.6J2 Both sides of a titanium sheet (0.7 cm X 0.7 cm X 0.05 cm) were coated with a W 0 3 layer of 10 pm. For some experiments one side was insulated with plastic paint. Ru0, electrodes were prepared according to Trasatti et al.I3 Finely divided W 0 3 suspensions were obtained by ultrasonification for 10 min. To
-
(9) Krasnovskii, A. A.; Brin, G. P. Dokl. Akad. Nauk. 1962, 147, 656. (10) Darwent, J. R.; Mills, A. J . Chem. SOC.,Faraday Tmns. 2 1978, 78, 359. (11) Ward, M. D.; Bard, A. J. J . Phys. Chem. 1982, 86, 3599. (12) Desilvestro, J.; GrHtzel, M. J . Chem. SOC.,Chem. Commun. 1982, 107. ( 1 3) Galizzioli, D.; Tantardini, F.; Trasatti, S. J . Appl. Electrochem. 1974, 4, 57.
0 1985 American Chemical Society
