5332
J. Phys. Chem. 1994,98, 5332-5337
Photoinduced Electron-Transfer Reactions between Dicyanobis(polypyridine)ruthenium(11) and Tris(&di ketonato)ruthenium(111) Complexes Mutsuhiro Maruyama, Noriyuki Sonoyama, and Youkoh Kaizu’ Department of Chemistry, Tokyo Institute of Technology, 0-okayama, Meguro- ku, Tokyo, Japan Received: October 6, 1993; In Final Form: March 16, 1994’
Photoinduced electron-transfer (ET) reactions of Ru(CN)2L2 (L = bpy and phen derivatives) with a series of RulI1 (@-diketonato)3were investigated in acetonitrile solutions and compared with organic quencher systems (TCNQ, benzoquinone, and nitrobenzene, etc.). In the normal and the plateau region (AG > -1.2 eV), the quenching rate constants of Ru(@-diketonato)3are smaller than that of organic molecules a t the same AG. The disadvantage of the ET of the Ru(@-diketonato)3 quenching system arises mainly from the low transmission coefficient caused from the weak orbital overlap between donor and acceptor. Although a slight decrease of quenching rate was found for each donor in the highly exothermic region (AG < -1.6 eV) in which Ru(hfac)3 was used as a quencher, evidence of the quadratic inverted region was not obtained. Some reasons for the weak AG dependence of k, in this region are discussed.
Introduction The electron-transfer (ET) reaction is one of the most fundamental processes in chemistry, and there has been much interest in the factors which govern the ET The electronic transmission coefficient determines the reaction adiabaticity. The experimental evaluation of the degree of the electronic exchange interaction has been widely performed using the intramolecular ET systems.’ The estimation of it also has been performed for the ~elf-exchange~ and the electrochemical ET system^.^ However, the survey for the bimolecular systemsG9 is limited. The degree of the electronic interaction, for bimolecular systems, is considered to be dependent on the orbital character of the redox center because it depends on the electronic (the electronic density4 and the electronic phaselo) and the steric factors. The r or r* orbitals of aromatic compounds are delocalized over the rings, while the d r orbital of metal complexes is metal-centered and it is surrounded by the ligands. Thus, in view of the orbital overlap, the metal complexes are considered to beat a disadvantage in comparison with the aromatic molecules. For example, the quenching through the exciplexlla or through the charge-transfer statellb has been observed for the reaction between aromatic compounds, while such a quenching mechanism has not been observed for metal complexes1* without a few exceptions.13 It is known that the strong interaction of theexciplex of the aromatic molecules comes from a parallel configuration of a close (3-4 A) interplanar ~eparati0n.l~ For metal complexes, such a close D-A configuration is not probable because of the steric hindrance of the ligands. The purpose of this paper is to answer the following questions. (1) To what extent does the orbital overlap differ between d?r and r orbitals as the accepting orbital? (2) How does such difference affect the AG dependence of the ET rates? Such comparison of the orbital overlap has not been done in detail up to this point. In this work, theETreactionsfromexcitedRu(L)2(CN)2 where L = 2,2’-bipyridine and 1,lO-phenanthrolinederivatives to a series of Ru(@-diketonat0)3(whose accepting orbital is metal d r ) or to a series of organic molecules such as TCNQ, benzoquinone, and nitroaromatics (whose accepting orbital is r*) are studied and their AG dependences on the quenching rate are compared. A series of Ru(&diketonato)3, chosen as the electron acceptors, have no low-lying d-d excited states which can be used in an energy-transfer reaction,15-17and they also have no electrostatic
* To whom correspondence should be addressed. 0
Abstract published in Advance ACS Abstracts, April 15, 1994.
0022-3654/94/2098-5332%04.50/0
effect such as ion-pairing, by which static quenching will occur, because they have no formal charge. Owing to the variety of reduction potentialsL9of these complexes, we could compare the AGdependenceof ET rates with that of organicquencher systems in a wide AG range. Materials Ru(4,4’-dmbpy)z(CN)2, Ru( 5,5’-dmbpy)z(CN)z, and Ru(4,7dpphen)z(CN)z, where 4,4’-dmbpy, 5,5’-dmbpy, and 4,7-dpphen denote 4,4’-dimethyL2,2’-bipyridine, 5,5’-dimethyl-2,2’-bipyridine, and 4,7-diphenyl-1,IO-phenanthroline, respectively, were prepared according to the literature method.20 Ru(4,4’-dmbpy)2(CN)2 and Ru(5,5’-dmbpy)z(CN)z were purified by column chromatography (Wakogel, Wako Co., Ltd.) in methanol and recrystallized from methanol/water. For R~(4,7-dpphen)~(CN)~, the separation of the product and the unreacted free ligands by column chromatography was unsuccessful. The free ligands were removed by column chromatography (Silica gel 40, Merck) as the ferrous chelates which were formed by adding the ferrous sulfate to the methanol solution of the crude products. Although a small amount of the free ligand was detected by the emission measurement after the recrystallization from acetonefhexane, further purification was not done. A series of Ru(&diketonato)3 quenchers, R ~ ( d p m )Ru(acp)3, ~, Ru(acac)3, Ru(dbm)3, R ~ ( t f a c )Ru(btfa),, ~, and Ru(hfac)~,were prepared according to the literature procedure.21 The structure of 8-diketonato is shown in Figure 1, and the abbreviations are summarized in Table 1. Ru(hfac)3 and Ru(acp)3 were purified by sublimation in vacuo, and others by column chromatography (aluminum oxide 90, Merck) in benzene. The organic quenchers purchased from Wako Pure Chemical Industries, Ltd. (TCNE and TCNB), Kanto Chemical Co., Inc. (TCNQ), and Tokyo Kasei Kogyo Co., Ltd. (others) were purified as follows. TCNE, TCNQ, BQCL, BQC12, and BQ were sublimed in vacuo. TCNB was used as supplied. Other samples were purified by the literature procedure.22 Tetraethylammonium perchlorate (TEAP) as a supporting electrolyte was prepared according to the literature23 and dried in vacuo for 20 h before use. The solvent acetonitrile from Wako was purified by distillation and used after the second distillation over CaH2. Apparatus and Measurements The second harmonics (532 nm) of the Nd3+:YAG laser (Spectron Laser Systems) was used to excite the donors. The 0 1994 American Chemical Society
Photoinduced Electron-Transfer Reactions
0
The Journal of Physical Chemistry, Vol. 98, No. 20, 1994 5333
0
Figure 1. Structure of the 8-diketonato ligand. The 8-substituents (R,, R2) are summarized in Table 1.
TABLE 1: Abbreviations for the Ligands of Ru(8-diketonato)3 complex Ru(hfac)3
ligand 1,1,1,5,5,5-hexafluoro2,4-pentanedionate
Ru(btfa)3
4,4,4-trifluoro-l-phenyl-
1,3-butanedionate Ru(tfac)3 Ru(dbm)a Ru(acac)3 Ru(acp)3 Ru(dpm)3
’Ph
&substituents (RI, Rz) (CF3, CF3) (CF3, Ph‘)
l,l,l-trifluoro-2,4-pentanedionate (CF3, CH3) 1,3-diphenyl-1,3-propanedionate (Ph, Ph)
2,4-pentanedionate 5,5-dimethyl-2,4-hexanedionate
2,2,6,6-tetramethyl3,Sheptanedionate CsH5. * BU = C4H9.
(CH3, CH3) (CH3, t-Bub) (t-Bu, t-Bu)
emission from the donor was collected in the photomultiplier (Hamamatsu R928) through the monochromator (Nikon P250). The time profile of the emission was recorded on the digital oscilloscope (Lecroy Model 9450) and was transferred and analyzed on the microcomputer (NEC PC-9801 VM). The sample solution was thermostated at 298 K by the Yamato NeoCool (Model CTE-21). The absorption and the emission spectra were determined on the Hitachi spectrophotometer Model 330 and on the Hitachi 850 fluorescence spectrophotometer equipped with Hamamatsu R928 photomultiplier, respectively. The emission spectra were corrected with the standard solution of 4-dimethylamino-4’-nitrostilbene.24 The cyclic voltammetry was carried out on the potentiostat (Hokuto Denko HA-501), equipped with the function generator (Hokuto Denko HB-104), and recorded on the X-Y recorder (Riken Denshi Model F-4). Thequenching rate constant (k,) in acetonitrile was determined by the conventional Stern-Volmer (SV) plot for the emission lifetime. For the organic quencher system, the emission intensity measurement also gave the same rate constants. For the Ru(@-diketonato)3 series, however, the intensity measurement resulted in the nonlinear SV plot because all the R~(@-diketonato)~ studied in this work absorb the excitation light in the visible and the attempt to correct such inner-filter effects resulted in failure. Sample solutions were freed from oxygen by bubbling of argon gas before use. The concentration of the R u ( L ) ~ ( C Nwas ) ~ about M, and that of the quencher was not above M. The cyclic voltammetry in acetonitrile was performed against the Ag(s)/AgN03 (0.1 M) in CH$N reference electrode with 0.1 M tetraethylammonium perchlorate as a supporting electrolyte. The concentration of the solute was below 10-3 M, and the scan rate was set at 200 mV/s. Results and Discussion Determination of 4.The lifetime of the luminescence of Ru(L)2(CN)2 decreased with the increase of the concentration of Ru(@-diketonato)3or organic quenchers. Mixed solutions of these compounds gave a superposed absorption spectra except for the case of Ru(hfac)3 in the concentration range used in this work. As is shown in Figure 2, the Stern-Volmer plot for the emission lifetime of Ru(L)2(CN)2 gave straight lines for each quencher. Obtained k, values are summarized in Table 2 and Table 3. For the case of R ~ ( h f a cas ) ~a quencher, thermal reduction of RIP’(hfac)g to Ru11(hfac)3- occurred.25 However, since the amounts of the reduced species after 1 h were estimated to be 0.5-1% and the lifetime of the solution at this time was the same as the freshly
00
0 5
1 0
1 5
2 0
2 5
30X1O4
concentration of the quencher (M)
Figure 2. Stern-Volmer plots ( T O / T vs concentration of the quencher) of Ru(4,7-dpphen)~(CN)2quenched by Ru(btfa), (A),Ru(hfac)p ( O ) , Ru(acac)3 (0),and Ru(acp)s (A).
TABLE 2: Rate Constants for Quenching of Ru(L)~(CN)~ by the Ru(B-diketonato)jSeries Ru(4,4’-dmbpy)2- Ru(5,5’-dmbpy)2- Ru(4,7-dpphen)z( W 2 ’ (CN)2’ (CN)zc 1 ~ 9 4 AG 1 ~ 9 9 AG ~ 10-9k AG quenchers (M-I 9-l) (eV) (M-I s- ) (eV) (M-l s-7) (eV) Ru(hfac)s 7.25 -1.78 8.89 -1.87 6.35 -1.69 (+0.29 V)’ Ru(btfa)3 9.57 -1.09 10.6 -1.17 9.76 -0.99 (-0.41 V)’
Ru(tfac)j 9.27 -0.98 10.4 -1.06 9.04 -0.88 (-0.49 V)‘ Ru(dbm)3 8.72 -0.53 9.78 -0.61 7.85 -0.43 (-0.91 V)’ Ru(acac)3 5.90 -0.28 7.25 -0.36 4.35 -0.18 (-1.41 V)’ -0.14 4.14 -0.22 1.08 -0.04 Ru(acp)3 3.20 (-1.34 V)’ Ru(dpm)l 0.48 +0.01 0.97 -0.07 d +0.11 (-1.56 V)’ E(D/D+) = 0.35 V vs 0.1 M Ag/AgNOs in acetonitrile; emission maximum = 13.9 X 103 cm-I; T = 210 ns in deoxidized acetonitrile at 298 K. E(D/D+) = 0.38 V; emission maximum = 14.7 X lo3 cm-I; T = 890 ns under the same conditions as a. E(D/D+) = 0.45 V; emission maximum = 14.0 X 103 cm-1; T = 1.7 ws under the same conditions as a. Could not be obtained because of the low solubility of Ru(dpm)o. The reduction potential E(A-/A) vs 0.1 M Ag/AgNOo in acetonitrile. prepared solution within the error of 1%, such thermal reduction is considered to have a negligible effect on determining k,. Estimation of the Free Energy (AC). Since the quenching rate constants decreased as the reduction potential of the acceptor became more negative, as shown in Table 2 and Table 3, the oxidative quenching is considered to occur in our systems26
D*
+ A-
D++ A -
(1)
where D* represents the excited donor, D+ is its oxidized form, and A and A- represent an acceptor and its reduced form, respectively. The free energy change (AG) of the “charge separation” E T reaction could be written as the follo~ing:2*,~~
AG = E(D/D+) - E(A-/A) -E,
e‘ -;
(2)
Here, E(D/D+) is the oxidation potential of a donor, E(A-/A) is the reduction potential of an acceptor, e is the charge of the electron, t is the static dielectric constant of the solvent, and r is the distance between the two reactants. The last term (- ez/tr) represents the static attraction energy of the products. The radii of Ru(L)z(CN)2, Ru(@-diketonato)3, and organic quenchers are estimated to be 7,30 7,32 and 3.8 A,31 respectively, leading to the intermolecular distance ( r ) of 14 A (for the Ru(@-diketonato)3,and 10.8 A (for the organic quencher system).
5334
The Journal of Physical Chemistry, Vol. 98, No. 20, 1994
Maruyama et al.
TABLE 3: Rate Constants for Quenching of Ru(L)z(CN)z by Organic Molecules quenchers TCNE (tetracyanoethylene) TCNQ (tetracyanoquinodimethane) BQCl4 (tetrachlorobenzoquinone) BQCl2 (dichlorobenzoquinone) BQ (benzoquinone) pyromellitic dianhydride p-xyloquinone 1,4-naphthoquinone TCNB (tetracyanobenzene) tetrachlorophthalic anhydride methyl p-nitrobenzoate methyl m-nitrobenzoate p-nitrotoluene p-nitroanisole
E(A-IA)a (V)
L = 5,5’-dmbpy lV9k,(M-’ s-’)/AG(eV)
-0.12
17.31-1.36
-0.18 -0.34 -0.45
16.61-1.30 15.01-1.14 15.01-1.03 13.71-0.62 10.2/-0.61 11.1/-0).48
-0.86
-0.87 -1
.oo
L = 4,4’-dmbpy lV9kq/AG
17.91-1.28 15.71-1.22 b b
11.31-0.54 8.631453 8.951440 9.361436 9.271-0.35
-1.04 11.31-0.44 -1.05 10.41443 -1.19 8.041429 b -1.30 8.06/-0.18 6.30/-0).10 -1.40 5.161-0.08 4.69/0.00 -1.57 2.02/+0.09 b -1.62 0.959/+0.14 b 2,5-dimethylnitrobenzene -1.65 0.555/+0.17 0.35/+0.25 The reduction potential of the quencher vs 0.1 M AgIAgNO3 in CH3CN. Not measured. The value of c in acetonitrile is taken from the reference as 37.5.34 Then the static attraction energy is evaluated to be -0.03 eV for both quenching systems. E m is the zero-zero excitation energy of the 3MLCT state of donors. In this paper, the energy of the emission maximum at room temperature is approximated as Em without further correction for a shift from the (0,O)band because the 3MLCT state of R u ( L ) ~ ( C Nis) ~too weak to be resolved in the absorption spectra and the emission at room temperature is broad and structureless. This approximation leads to the values of 1.72, 1.83, and 1.73 eV for Ru(4,4’-dmbpy)2(CN)2, Ru(5,5’-dmbpy)2(CN)2, and Ru(4,7-dpphen)2(CN)2, respectively. Estimated AG : values are summarized in Table 2 and Table 3. 2 ’ I I I -2 0 .l 5 -1 0 -0 5 00 The obtained Emvalues from the emission maximum a t room AG (eV) temperature may include some systematic errors, while the Figure 3. Plot of log k, vs AG for *Ru(L)z(CN)2/Ru(@-diketonato)3 magnitude of the discrepancy is estimated to be around 0.05eV. systems ( ( 0 )L = 5,5’-dmbpy, (A)L = 4,4’-dmbpy, (m) L = 4,7-dpphen) Several efforts have attempted to estimate the Em values of and for *Ru(L)2(CN)2/organicquencher systems ((0)L = 5,5’-dmbpy, Ru(L)2(CN)2. Demas et used the energy at which the (A)L = 4,4’-dmbpy). The brokencurvesrepresentthe best-fit correlation emission intensity, in the short wavelength side, falls to 5% of the of the data in the normal region with the assumption of G(O)*i, = 0 eV: intensity at the emission maximum at 77 K. Kitamura et al.,36 A = 1.3 X 1Olo (for the Ru(fl-diketonato)3system) and A = 3.0 X 1013 however, pointedout that thegreat changes ofthesolvation around (for the organic quencher system). For other parameters, see text. the complex occur at the glass transition temperature of the medium and that the Em values estimated from the emission at D + A kd t Da#**awan*ma 77 K do not necessarily represent those at room temperature. k-d k.e D+A Thus, they estimated Em as the slightly (ca. 0.05 eV) large value Figure 4. Kinetic scheme of the quenching reaction. D and A represent of the emission maximum at room temperature. a donor and an acceptor, respectively. Nagle et al.37estimated the oxidation potential of the 3MLCT state of Ru(bpy)32+ by applying the E T theory to the quenching the scheme shown in Figure 4,6-22929938 where kd is the diffusion experiment and found that the obtained E m is slightly larger rate constant, k 4 is the dissociation rate constant, k, and k, are than the energy of the emission maximum at room temperature the forward and backward electron-transfer rate constants, and by the factor of 0.05 eV. They used the value of 0.05 eV as the k, is the rate constant for the radical ion dissociation and/or the correction factor for the estimation of E m from the emission of charge recombination. Under the assumption that the k, is much Ru(L)2(CN)2 at room temperature. larger than k, the overall quenching rate constant is written as Due to the fact that the structures of donors used in this work eq 3.29,38With a classical treatment,2,6 using Marcus function,’ resemble each other and the Stokes shifts are likely to be quite similar, our estimation of Em might lead to a systematic negative kd k, = (3) shift of AG as large as 0.05 eV a t the maximum. k4 The Relationship between log 4 and AG. In Figure 3, log k, 1 ke is plotted against AG. For the Ru(P-diketonat0)3 series, the rate constants decrease as the driving force (-AG) decreases in the kq is rewritten as the function of AG as eq 4, where K is the region of -AG < 0.7 eV (the normal region). At 0.7 eV < -AG < 1.2 eV, the rate constants remain almost constant at the value kd 1.1 X 1010 M-1 s-I (the plateau region). In the normal and the k, = (4) plateau regions, the quenching rate constants of the Ru(Bdiketonato)3 system are found to be smaller than those of the 4AG(O)* organic quencher system when compared a t the same AG value. Although a slight decrease of the rate constant is observed for transmissioncoefficient of theelectron transfer, Y is the frequency each donor (Table 2) in the highly exothermic region (-AG > 1.6 factor, Kd is the equilibrium constant of the encounter complex eV), in which Ru(hfac)3 is used as the quencher, these points (Kd = kd/kA),and AG*(O) is the activation free energy when AG rather scatter and the clear inverted region is not observed. equals 0 eV (intrinsic barrier). Here, AG*(O) is considered to Evaluation of the Parameters of the ET Rate. The quenching be the sum of the inner-sphere (AG*(O)i,) and the solvent reaction of the excited R u ( L ) ~ ( C N in ) ~solution is described as (AG*(O),,,) activation energy (eq 5). , I
+-
Photoinduced Electron-Transfer Reactions
The Journal of Physical Chemistry, Vol. 98, No. 20, 1994 5335
Equation 4 shows that the kq value is strongly dependent on the intrinsic barrier AG*(O) and the preexponential factor K V . Since the quadratic inverted region was not observd for our quenching systems, the evaluation of the parameters of the ET is attempted in the normal region (AG > -0.7 eV). We used the following assumptions for v , AG*(0),,t, kd, and k d in this work. The frequency factor (v) depends on the frequency of the vibrational modes, intramolecular or solvent, which contribute to the destruction of the activated encounter complex. According to Sutiqza it varies from the order of 10l2 to 1014 s-1. In this work, v is approximated on the order of l O l 3 s-l 39 for both quenching systems. It should be noted that a different mode may contribute to the deactivation for the two quenching systems studied here. However, it is difficult to estimate the concerted vibrational frequencies for each case. The solvent activation energy (AG*(O),J is estimated by1~40
AG*(O),,, =
"'( 1 ')( 2 1 - --) 1 4 2r, + 2rA a -
(6)
where r D and r A are the radii of a donor and an acceptor, a is the distance between a donor and an acceptor, and n is the refractive index of the solvent. The value of n of acetonitrile a t 298 K is 1.34.34 Then, AG*(O),,, is estimated as 0.139 and 0.215 eV for Ru(@-diket~nato)~ and organic quencher systems, respectively. The diffusion rate constant (kd) is approximated by the experimental plateau value, leading to 1.1 X 1010 and 1.5 X 1010 M-l s-l for the Ru(@-diketonato)3 and the organic quencher systems, respectively. It should be noted that the plateau value is slightly decreased by the weak nonadiabaticity ( K = 10-2l e 3 ) . As is mentioned in the Appendix, the plateau value becomes about 10% smaller than kd if K is as large as le3. However, this approximation for kd has a negligible effect on the calculation in the normal region. The equilibrium constant of the encounter complex (&) is represented as eq 7, following Fuoss and Eigen,4' where N A is Avogadro's number, leading to the values of 6.62 and 3.18 M-I
(7) for the Ru(@-diketonato)3system and for the organic quencher system, respectively. By using the Kd and kd values, the dissociation rate constants ( k d ) are estimated as 1.53 X 109 and 4.72 X 109 s-l for each system. For the calculations in which AG*(O)i, and the electronic factor KV (EA ) are adjusted to fit the experimental results, the parameter set (AG*(O)i,, A ) = (0 eV, 3 X 10") reproduced the results for organic quenching systems. Other parameter sets such as (0,OS eV, 3 X 1014)also reproduced it. However, this set is unreasonable because the order of 1014of A is considered to be too large. Further large values of A are needed to fit the result for further large values of AC*(O)i,. For Ru(@-diketonato)3quenching systems, some parameter sets reproduce the experimental result. For example, the parameter sets (AG*(O)i,, A ) = (0 eV, 1.3 X lolo), (0.1 1 eV, 9.3 X loll), and (0.17 eV, 1 X 1013)reproduce it; both AG*(O)i, and A vary in a wide range. AG*(O)i, for the quenching reaction between the excited RuII-bipyridine complex and the organic molecule has been frequently assumed to be negligible,6J while little is known about the AG*(O)i, for the reduction process of Ru(@-diketonat0)3.~From the viewpoint of the low spin (d6-d5) electronic configurations of RuI1-RulI1 redox systems, AG*(O)i, is considered to be negligible. As was pointed out for the R u ( N H ~ ) ~ ~redox + / ~ system + by Stynes and Ibers,42 however, the electrostatic interaction between the metal and the ligand will change the metal-ligand distance during such redox reactions. The significant bond length change of Ru-0 was observed by
Bernhard et a1.43 for the R ~ ( H 2 0 ) 6 ' + / ~redox + system. Using the difference of the metal-ligand distances between the oxidized (Ru(II1)) and reduced (Ru(I1)) form (Ar) obtained from the crystallographic data, they estimated AG*(O)i, of R U ( H Z O ) ~ ~ + / ~ + and R u ( N H ~ ) ~ ~as+ /0.1 ~ +and 0.025 eV, respectively. Unfortunately, the crystallographic data of the Ru(@-diketonat0)system are lacking, and the degree of AG*(O)i, for the Ru(@diketonat0)3~/-system cannot be evaluated here. In the present situation, it is useful to examine the limiting case; the value of AG*(O),, comes to its maximum when the A is at a maximum (adiabatic limit, K = 1). The parameter set of (0.17 eV, 1 X 1013) is this case. Compared with the evaluated AG*(O),, values of R ~ ( H 2 0 ) 6 ~ +(0.1 / ~ +e v ) and R U ( N H ~ ) ~ ~ + / ~ + (0.025 eV), thevalueof0.17 eV fortheRu(@-diketonato)Jsystem seems to be too large. Thus, it is necessary to introduce some nonadiabaticity for this quenching system. The fitting curves shown in Figure 3 are obtained by assuming AG*(O),, to bezero for both quenchingsystems. The experimental results are reproduced when the electronic factor A is 1.3 X 1010 for the Ru(@-diketonato)3and 3.0 X 1OI3for theorganicquencher systems, respectively. Using the order of l o L 3s-1 for v, K is estimated to be of the order of le3for R~(@-diketonato)~ and 1 for organic quencher systems, respectively. This result indicates an adiabatic and a nonadiabatic ET mechanism for the organic and the R~(@-diketonato)~ quenching systems, respectively. The adiabaticity of the E T reaction is influenced by the extent of spatial orbital overlap between a donor and an acceptor. The redox center of the Ru(II1) complexes (da) is metal centered, while that of organic quenchers (a*)is delocalized over the rings. It is expected that the orbital overlap of the Ru(@-diketonato)3 system is poor and the E T proceeds via the nonadiabatic mechanism. It should be mentioned that the metal d a electron delocalizes over the O-diketonato a orbitals" and the delocalization of the d a into the ligand a orbitals should favor the interaction with the donor. According to the modified extended Hiickel calculations,45 the LUMO of a series of R~(@-diketonato)~ complexes has 2030% ligand a character. However, such extent of delocalization will still provide poor interaction with a donor in comparison with the organic molecules whose LUMO has 100% a* character, because the degree of the interaction is considered to be proportional to the electronic population on the liga11d.~6Nielson et aL4observed 10-fold larger rate constants for the homogeneous self-exchange reactions of CpZFe+lo(ferrocenium-ferrocene) than for those of the CpzCo+/O (cobaltocenium-cobaltocene) couple, which could be attributed to the greater (45-50%) liganddelocalized character of the HOMO of Cp2Fe+ as compared with that of Cp2Co+ (10-14% ligand character). It is concluded that, even though the d a electron delocalizes over the ligand, the interaction of Ru(@-diketonato)3is still smaller than that oforganic quenchers. Weaver et aL5 also reported that the rate of the electrochemical exchange E T reaction of R ~ ( h f a c was ) ~ small and insensitive to the solvent dynamics compared with organometallic compounds, which indicated the nonadiabatic ET mechanism for this species. This nonadiabatic ET behavior of R ~ ( h f a c at ) ~the electrode is consistent with our result that the quenching of the excited donors by the Ru(@-diketonat0)3series proceeds via a nonadiabatic E T mechanism. It also should be mentioned that the sizes of the @-substituents (-CF3, -D6H5, -tert-C4Hg, and -CH3) of Ru(P-diketonato), in this workdiffer each other, and each ,&substituent may lower the orbital overlap to a different extent. Wilkinson and Tsiamis17a have reported that the steric reduction in the transmission coefficient by replacing the methyl groups of Cr(acac)j with tertbutyl groups was a factor of 7.2-62 for the energy-transfer reactions from the triplet aromatics such as anthracene, acridine, and pyrene. However, it is expected that, as the molecular size of a donor increases and the distance between the donor and the acceptor becomes larger, the difference in the steric factor of @-substituentswill become smaller, because the difference in the
5336 The Journal of Physical Chemistry, Vol. 98, No. 20, 1994
5-
-
4
7
D
k i
-3
m
3-
.
A
”
m
2-•
T
m
2
lo*;
A-
I
I
I
I
L-
Maruyama et al. the exciplex:2 long-range electron transfer,”b+s3quantum effect,54 and the dielectric saturation of the polar solvent molecule around the charged molecule produced by ET.” The exciplex formation will lower the effective AG by the strong electronic mixing between D and A. However, taking into account that the redox orbitals of Ru(&diketonato)3 are shielded by the ligands, the electronic mixing with the donor will not be adequate to form the exciplexes for Ru(&diketonato)3 redox systems. So this explanation seems unreasonable in this case. The ET at a long-distance separation shifts the maximum of the AG dependence of ET rates to the large -AG region because of the -1 / r dependence of the AG*(O),,,. For example, when the ET occurs at a = 2(rD + r A ) instead of a = r~ r ~the, maximum of the AG dependence of E T rates shifts from -0.554 to -0.831 eV. Thus, the long-distance E T will be favorable in the highly exothermic region if the nonadiabaticity is not so large a t the larger u. During the last decade, the quadratic inverted region has been observed for the D-(spacer)-A systems56and geminate radical pair systemsS7in which the D-A is fixed or in close contact. These observations indicate that the long-distance ET is the possible reason for the lack of the inverted behavior for such intermolecular E T systems as ours. Although the roles of the quantum effect and the dielectric saturation of the solvent are not clear in this work, it is concluded that the existence of the low-lying excited states and of various vibrational modes, in addition to the distribution effect between D and A, will provide them with favorable quenching pathways in the highly exergonic region. To know the detailed quenching mechanism of Ru(II1) complexes in this region, more quenching experiments are needed.
+
size of the &substituents is regarded as small at large D-A distances. Thus, the steric difference in the transmission coefficient between -CH3 and -tert-C4H9 of 8-substituents in this work is considered to be smaller than the reported value of 7.2-62 for the organic donor system. Anyway, such extent of steric difference is considered small compared with lo3, which is the estimated reduction value in the transmission coefficient for the Ru(L)2(CN)z/Ru(&diketonato)3 system in comparison with the R~(L)~(CN)2/organic quencher system. It is concluded that although each @-substituentmay lower the orbital overlap to a different extent, the poor orbital overlap due to the metal-centered orbital character of Ru(&diketonato)3 is the main factor for the lower transmission coefficient for RuSummary (@-diketonato)3than for organic quenchers. Photoinduced electron-transfer reactions between the Ru(L)2Explanations for the Weak AG Dependence of the Quenching (CN)2 and R~(@-diketonato)~ series were investigated in a wide Rates for Ru(&diketonato)3 Systems. Creutz and s ~ t i n ~ ~ ( C S ) AG region and compared with those of the organic quencher had performed the quenching experiments for the *Ru(L)s2+systems. The quenching rate costants of the Ru(j3-diketonato)3 R u ( L ) ~ ~O+ S, ( L ) ~ ~and + , C T ( L ) ~redox ~ + systems where L = bpy system were smaller than those of the organic quencher system or 4,4’-dmbpy. In the highly exergonic region, they observed the in the normal and the plateau regions at the same AG. The “vestiges” of the decrease, not the quadratic decrease. For the transmission coefficients of l e 3 and 1 were obtained for the RuRu(@-diketonato)3 system studied here, we also could not find (@-diketonato)3and the organic quencher systems, respectively, quadratic inverted behavior in that region in which Ru(hfac)3 by the curve-fitting analysis which assumed AG*(O)i, to be zero was used as the quencher, although a slight decrease of the k, for both systems. The low transmission coefficient for the Ruvalue was observed for each donor. In Figure 5, the results of (P-diket~nato)~ system will be attributed to the poor orbital overlap CS and our study are plotted together. Such a weak AG caused by the metal-centered character shielded by the ligands. dependence of the quenching rates in the highly exergonic region AG*(O)i, may not be zero, and the rate diminuation for the comes from several sources.48.49 Ru(&diketonato)3 system may come from the inner-sphere Siders and Marcus49’J pointed out that the formation of the reorganization. Unfortunately, we could not estimate the electronically excited products was the possible mechanism for appropriate value for it. Our calculations showed that some the cases of Ru(II1) as the quenchers of the CS reaction. If the nonadiabaticity is required for the Ru(j3-diketonato)~system to low-lying excited states exist in D+ or A-, the effective driving fit the experiment as long as AG*(O)i, is less than 0.17 eV. In the force (-AG) to form the electronically excited radical (A-* or highly exothermic region for Ru(B-diket~nato)~ quenching D+*) is reduced by its excited-state energy and thus the AG systems, the quadratic decrease was not observed. The exciplex dependence of the quenching rate constants will be weakened. formation will be excluded for the explanation of the weak AG As for the pathway to form (D+ A-*)50 in this work, the dependence of the quenching rates due to the small orbital overlap shoulderbandaround 12.5 X lO3cm-1(1.56eV) to 16 X lO3cm-l betweenDandAinthiscase. For theRu(hfac),case, thealternate (2.0 eV) of R ~ ( h f a c ) ~implies the existence of the low-energy pathway to form the electronically excited state of Ru(hfac)3- is excited states to which the E T may occur. Provided that the considered to be the probable mechanism. The long-distance E T excited-state energy is 1.6 eV, the effective AG in this case could is another possibility for the weak AGdependenceof thequenching be estimated as -0.09 to -0.27 eV. Thus, the effective driving rates. Although one can give other reasons such as quantum force to form the excited Ru(hfac)3- will come in the normal or effect and dielectric saturation of the solvents, these are not clear the plateau region, which implies the weak AG dependence of the in this work. quenching rates in the highly exergonic region. , For Ru(btfa), and Ru(tfac)3 in the plateau region, however, Appendix there is no low-lying excited state in Ru(btfa),- and Ru(tfac)3below 1.O eV and thus the formation of the electronically excited The Nonadiabatic Effect on the Plateau Region of the Ru(& products is considered to be the endothermic reaction. So, this diketonato)3 System. The small plateau value (1.1 X 1010 M-1 mechanism seems unreasonable in these cases, and other s-1) of the Ru(B-diketonat~)~ quenching system compared with explanations for the lack of decrease in this region are needed. that of the organic one (1.5 X 1Olo M-l s-l) indicates the Some reasons for the lack of quadratic inverted region have nonadiabatic ET for the former. On the basis of the diffusion been proposed, such as the quenching through the formation of theory in fluid solutions,58 the difference of the plateau value
+
The Journal of Physical Chemistry, Vol. 98, No. 20, 1994 5337
Photoinduced Electron-Transfer Reactions between two quenching systems seems large. According to S m o l u c h ~ w s k ikd, ~for ~ ~noncharged molecules is represented as eq 8, where q is the viscosity of the solvent. Using the values of
r D = 7 A, rA = 7 A (for Ru(&diketonato)3), and r A = 3.8 A (for organic quenchers), the kd value of the Ru(&diketonato)3 system is only 10% smaller than that of the organic quencher system. When the ET is nonadiabatic ( K is 10-2-10-3), the plateau value is slightly lower than kd. In the plateau region where the activation energy (AG*)is very small or zero, k, is represented as eq 9.59 Using the values of loi3s-l for v, 6.62 M-l for Kd, lolo kd
k, = 1
kd +-KVKd
(9)
’’
M-’ S-’ for kd, and 10-2-10-3 for K , kd/KVKd is derived as around0.015(~=10-2)and0.15 ( K = Finally,thefollowing relationship is obtained: k, = 0.99kd (when K = and k, = 0.87kd (when K = l e 3 ) . In the case of K = l e 3 , the plateau value is 13% smaller than kd. Consequently, in the plateau region, the smaller value of the Ru(@-diketonato), system compared with the organic quencher system comes not only from the large molecular size of Ru(Pdiketonato)3 but also from the nonadiabatic effect (as large as 10-3) of the ET. References and Notes (1) Marcus, R. A. J. Chem. SOC.1956, 24,966. (2) (a) Sutin, N. Acc. Chem. Res. 1982, 15, 275. (b) Sutin, N. Prog. Inorg. Chem. 1983, 30,441. (3) For example: Closs, G. L.; Calcaterra, L. T.; Green, N. J.; Penfield, K. W.; Miller, J. R. J. Phys. Chem. 1986, 96, 3673. (4) For example: Nielson, R. M.; Golovin, M. N.; McManis, G. E.; Weaver, M. J. J. Am. Chem. SOC.1988, 110, 1745. (5) Weaver, M. J.; Phelps, D. K.; Nielson, R. M.; Golovin, M. N.; McManis, G. E. J. Phys. Chem. 1990, 94, 2949. (6) Sandrini, D.; Maestri, M.; Belser, P.; von Zelewsky, A.; Balzani, V. J . Phys. Chem. 1985,89, 3675. (7) Kitamura, N.; Rajagopal, S.;Tazuke, S. J. Phys. Chem. 1987, 91, 3767. (8) Koval, C. A.; Ketterer, M. E.; Reidsema, C. M. J. Phys. Chem. 1986, 90, 420 1. (9) Berkoff, R.; Krist, K.; Gafney, H. D. Inorg. Chem. 1980, 19, 1 . (10) Gould, I. R.; Ege, D.; Moser, J. E.; Farid, S. J. Am. Chem. SOC.1990, 112, 4290. ( l ! ) (a) Kikuchi, K.; Niwa, T.; Takahashi, Y.; Ikeda, H.; Miyashi, T.; Hoshi, M. Chem. Phys. Le#.1990,173,421. (b) Foll, R. E.; Kramer, H. E. A; Steiner, U. E. J. Phys. Chem. 1990, 94, 2476. (12) For example: Shioyama, H.; Masuhara, H.; Mataga, N. Chem.Phys. Lett. 1982, 30, 161. (13) Dtacy, E. M.; McMillin, D. R. Inorg. Chem. 1990, 29, 393 and references cited therein. (14) Mattes, S. L.; Farid, S. Science 1984, 226, 917. (15) Gamache, R. E., Jr.; Rader, R. A.; McMillin, D. R. J. Am. Chem. SOC.1985, 107, 1141. (16) Huang, S.-M. Y.; Gafney, H. D. J. Phys. Chem. 1979, 83, 1902. (17) (a) Wilkinson, F.; Farmilo, A. J. Chem. Soc., Faraday Trans. 2 1981, 77, 1681. (b) Wilkinson, F.; Farmilo, A. Ibid. 1976, 72, 604. (18) Newsham, M. D.; Cukier, R. I.;Nocera, D. G. J. Phys. Chem. 1991, 95, 9660. (19) Patterson, G. S.; Holm, R. H. Inorg. Chem. 1972, 11, 2285. (20) Demas, J. N.; Turner, T. F.; Crosby, G. A. Inorg. Chem. 1969,8,674. (21) Endo, A.; Kajitani, M.; Mukaida, M.; Shimizu, K.; Sat8, G. P. Inorg. Chim. Acta 1988, 150, 25. (22) Kitamura, N.; Obata, R.; Kim, H.-B.; Tazuke, S. J. Phys. Chem. 1989, 93, 5764. (23) House, H. 0.;Feng, E.; Peet, N. P. J. Org. Chem. 1971,36, 2732. (24) Lippert, E.; Nagele, W.; Blankenstein, I. S.; Staiger, U.; Voss, W. 2.Anal. Chem. 1959, 170, 1 . (25) For the samplesolution containing 3.1 X l e M of Ru(4,7-dpphen)2(CN)2 and 3.1 X 10-4 M of Ru(hfac)~,the absorption band around 530 nm was slightly blue-shifted and increased in intensity with time. In addition, the absorptionband at 374 nmslightly decreasedin intensitywith time. Comparing the wavelength of the absorption maxima of Ru111(hfac), with that of KRuII( h f a ~ (see ) ~ below) reveals that the slight blue shift and rise around 530 nm
and the fall at 374 nm could be attributed to the reduction of R ~ ~ ~ I ( h ftoa c ) ~ Ru11(hfac)3-. The wavelength of the absorption maxima (nm) and molecular absorptioncoefficients (log(c/mol-I dm3 cm-1)) of UV-vis spectra in acetonitrile are as follows. Ru(hfach: 531 nm (loa e = 3.27). 374 (loa c = 3.94). 285 (log c = 4.09). KRu(hfac)l: 529 nm (hg c = 4.22), 494 (A),288 (log c = 4.31). 234 (1011 c = 3.981.21 (26) In ourquenching system, the energy-transfer mechanism to form the excited Ru(B-diketonato)3will be excluded sincesuch reaction is endothermic. Although the position of the d-d excited state of Ru(&diketonato), to which energy transfer may occur is obscured by strong CT bands, the energy of the lowest d-d state could be estimated around 2.0 eV from the reported values of the 0-coordinated complexes such as Ru(H20)63+ (2.06 eV)z7and R U ( O X ) ~ ~ (1.98 eV).27 Since the energy of the )MLCT state of R u ( L ) ~ ( C Nis) ~1.731.83 eV, the energy transfer to the lowest d-d state is endothermic and is considered to have a negligible effect in this case. Wilkinson and Farmilo17b suggested that the lowest excited state of Ru(acac), is the CT state which is located 2.1-2.5 eV above the ground state. They studied the energy-transfer reaction from the organic triplet donors to Ru(acac)3 and found that the threshold energies of the triplet state of organic donors for the energy transfer toRu(acac)s werearound2eV. Thek,valueofthe triplet 1,2-benzanthracene (whose triplet energy is 2.06 eV) is around 2.2 X lo9 (M-l s-l), while that of acridine (1.98 eV) is around 107-108 (M-I 8 1 ) . The k, value of anthracene (1.83 eV) is much smaller (