2892
J . Phys. Chem. 1992, 96, 2892-2901
Triplet-State Electron Transfer from Anthracene and Pyrene Covalently Bound to Polyelectrolytes in Aqueous Solution Jiunn-Shyong Hsiao and S. E. Webber* Department of Chemistry and Biochemistry and Center for Polymer Research, The University of Texas at Austin, Austin, Texas 78712 (Received: October 16, 1991; In Final Form: December 10, 1991)
The electron-transfer quenching of the triplet state of an aromatic chromophore (anthracene, pyrene) covalently bound to polyacids or polystyrenesulfonate has been studied. The quencher is a zwitterionic viologen (4,4’-bipyridinio-l, 1’-bis(propanesulfonate)) that becomes anionic upon reduction. The efficiency of charge separation was found to be largest (ca.0.8) at high pH for the anthracene-tagged polymer, in contrast with earlier results for the singlet state ( J . Phys. Chem. 1989, 93, 1928). For pyrene a number of different polymer backbones were studied, and it was found that the state of ionization of polyacids had little effect on the yield of ion pairs. (The yield was between 0.4 and 1.0, depending on the polymer.) This is also different than the singlet-state results, where no charge separation occurs at any pH ( J . Phys. Chem. 1988, 92, 2934). It is proposed that the local polymer structure has a specificity for the covalently bound chromophore, which in turn modifies the quenching mechanism. In all cases the ion pairs have a long lifetime, in excess of 2 ms. The T-T extinction coefficients were determined for polymer-bound anthracene and pyrene and were found to be much smaller than the unsubstituted species in organic solvents. The conditions for photoionization in these systems are also discussed.
Introduction Electron transfer from excited-state species has been actively studied in recent years for a large variety of chromophores in many different kinds of organized structures.’ We and others have explored the use of chromophores covalently bound to polyelectrolytes or other water-soluble polymers.* As was recognized many years ago by Morawetz and c o - ~ o r k e r sand , ~ emphasized more recently by Rabani et al.,” the charge density of a polyelectrolyte can be exploited to concentrate reagents and/or provide electrostatic repulsion of products in such a way that the overall yield of a reaction can be enhanced. Morishima et al.5 and we6 have also pointed out that amphiphilic polymers can modify the yield and kinetics of excited-state electron-transfer reactions by steric effects which arise from hydrophobic interactions between the polymer and chromophore (termed “compartmentalization” by Morishima and “hydrophobic protection” by us). Guillet and co-workers have used random copolymers composed of the hydrophobic chromophore with partial sulfonation to impart water solubility to effect sensitized photochemical reactions in hydrophobes (denoted “photozymes”) which make use of these same principles.’ (1) (a) See the plenary lectures of the Sixth International Conference on Photochemical Conversion and Storage of Solar Energy, Paris, 1986, in: New J. Chem. 1987, 11. (b) Fox, M. A., Chanon, M., Eds. Photoinduced Electron Transfer; Elsevier Science Publishers: Amsterdam, 1989. (2) For a review see: Rabani, J. In ref 1, Part B, Section 3.4. (3) (a) Taha, I.; Morawetz, H. J . Polym. Sci., Part A-2 1971, 9, 1669. (b) Taha, I.; Morawetz, H. J . Am. Chem. SOC.1971, 93, 829. (c) Vogel, B.; Morawetz, H. J . Am. Chem. SOC.1968, 90,1368. (d) Morawetz, H.; Gordimer, G. J . Am. Chem. SOC.1970, 92, 7532. (4) (a) Slama-Schwok, A.; Rabani, J. Macromolecules 1988,21,764. (b) Sassoon, R. E.; Rabani, J. J . Phys. Chem. 1985, 89, 5500. (c) Rabani, J.; Sassoon, R. E. J . Photochem. 1985, 29, 7. (d) Sassoon, R. E.; Aizenshatat, Z.; Rabani, J. J . Phys. Chem. 1985, 89, 1182. (e) Sassoon, R. E.; Rabani, J. Isr. J . Chem. 1982, 22, 138. ( 5 ) (a) For a recent review see: Morishima, Y . Prog. Polym. Sci. 1990, 15,949. (b) Morishima, Y.;Furui, T.; Nozakura, S.-I.; Okada, T.; Mataga, N. J . Phys. Chem. 1989,93, 1643. (c) Morishima, Y.; Kobayashi, T.; Furui, T.; Nozakura, S.-I. Macromolecules 1987, 20, 1707. (d) Itoh, Y.; Morishima, Y.; Nozakura, %-I. Phorochem. Photobiol. 1984, 39, 451. (6) (a) Chatterjee, P. K.; Kamioka, K.; Batteas, J. D.; Webber, S. E. J . Phys. Chem. 1991,95,960. (b) Stramel, R. D.; Webber, S. E.; Rodgers, M. A. J. J . Phys. Chem. 1989, 93, 1928. (c) Delaire, J. A,; Sanquer-Barriert, M.; Webber, S. E. J . Phys. Chem. 1988, 92, 1252. (d) Stramel, R. D.; Webber, S.E.; Rodgers, M. A. J. J. Phys. Chem. 1988,92,6625. (e) Stramel, R. D.; Nguyen, C.; Webber, S. E.; Rodgers, M. A. J. J. Phys. Chem. 1988, 9292934. (7) Nowalowska, M.; Guillet, J. E. Macromolecules 1991, 24, 474 and references therein to earlier papers.
0022-3654 192/2096-2892$03.00/0
SPV, 6
PMA.A, 1
COOH
COOH
PMA-vPy, 1
PNP
Q
In this paper we present a study of the triplet state of anthracene and pyrene covalently bound to water-soluble polymers (see Chart I) which represent a number of different hydrophobic and/or charge densities. The fundamental reaction can be represented ?.s+lArl + V
-
-
2s+l{Ar.+,V*-l k,
km
?.Ar*++ 2V.-
-Ar+V with a charge separation quantum yield given by
(la) (Ib)
(2) = k , / ( k , + krm) In the above V represents a general electron acceptor, which in much work has been a viologen. In the present case we use the zwitterionic viologen 4,4‘-bipyridinio- 1,1’-bis(propanesulfonate), abbreviated SPV (see Chart I). In earlier work we found that dJcs
0 1992 American Chemical Society
Electron-Transfer Quenching in Polymers
for poly[(methacrylic acid)-co-vinylpyrene] (PMA-vPy) that no charge separation occurred upon quenching the singlet state with viologen species at any pH,6e while for the corresponding anthracene polymer a charge separation yield (&) of ca. 0.4 occurred at low pH.6b For the triplet excited state and all polymer backbones Vy*produces a significant yield of long-lived (>2 ms) ion pairs, while for PMA-jA* efficient quenching and charge separation occur only a t high pH. Thus, the need for “compartmentalization” is diminished for triplet states because of the spin forbiddeness of the recombination reaction (eq 1b), and the steric requirements for charge separation following the quenching of excited states can be different for triplet and singlet excited states. Experimental Section Preparation of Polymers. The anthracene- and pyrene-containing polymers (1,2,3, and 5 in Chart I) were prepared by Dr. R. D. Stramel using standard free-radical polymerization methodS.6C
Copolymers of styrene and maleic anhydride with vinylpyrene (PMS-vPy, 4) were prepared as follows. Styrene (Aldrich) was distilled under vacuum prior to use. Vinylpyrene was synthesized according to the procedure of Tanikawa et a1.* Maleic anhydride (Aldrich) was recrystallized three times in chloroform and sublimed under vacuum twice prior to use. 2,2’-Azobis(isobutyronitrile) (AIBN, Kodak) was recrystallized twice from methanol. Benzene (J.T. Baker) was freshly distilled. Styrene (5.45 mmol), maleic anhydride (40 mmol), and vinylpyrene (0.05 mmol) were copolymerized in benzene with AIBN (0.14 ”01) as an initiator. The mixture was degassed by several freezepumpthaw cycles, and the polymerization was carried out at 60 OC for 30 min. The polymers were purified by repeated precipitation of an acetone solution by addition to excess 9: 1 hexane/benzene cold solvent. The copolymer was hydrolyzed by treating a highly concentrated T H F solution with 0.25 M sodium hydroxide. Typically, ca. 0.5 g of polymer in ca. 1 mL of T H F was mixed with 100 mL of sodium hydroxide solution and heated at 50 O C for 3 days. Next, the solution was dialyzed against pure distilled water for several days, changing the water frequently, and finally lyophilized to yield an off-white fluffy solid. Esterification of the polymer for characterization was carried out as follows. A basic aqueous solution of the hydrolyzed polymer was acidified with an excess of concentrated HCl, and the precipitated polymer was collected by centrifugation. (The polymer tends to aggregate for pH In order to determine whether the copolymer propagates via an alternating sequence distribution, UV absorption and fluorescence techniques were applied. The UV absorption showed that the copolymer tended toward a 1:l composition. Also, as mentioned in the Experimental Section, fluorescence studies revealed almost no styrene-styrene excimer band for the model polymer PMS, which suggests that there are few styrene neighbors in the polymer backbone. ps, respectively.
( 1 3) Watanabe, T.;Honda, K.' J. Phys. Chem. 1982,86,2617. The valucs given are for MV'+ (methylviologen) and have been used for S P V (Willner, I.; Yang, J.; Laane, C.; Otvos, J.; Calvin, M. J . Phys. Chem. 1981, 85, 3277). (14) (a) Lappalainen, E.; Koskimies, S . J . Po/ym. Sci., Polym. Left. Ed. 1986, 24, 17. (b) Trivedi, B. C.; Culbertson, B . M. Maleic Anhydride; Plenum Press: New York, 1982; p 307. (c) Cowie, J. M. G. Alternating Copolymers; Plenum Press: New York, 1985. (d) Hill, D. J. T.;ODonnell, J. M.; 0'Sullivan, P. W. Macromolecules 1985, 18, 9.
The Journal of Physical Chemistry, Vol. 96, No. 7, 1992 2891
Electron-Transfer Quenching in Polymers 1
I
360
560
460
-251 310
760
660
(-1 Figure 7. Spectral components for PSS-vPy. I
I80,
.
.
"
"
360
410
.
x (-1
'
460
.
'
510
.
' 1 560
Figure 9. S-S and T-T absorption for PMA-vPy at pH 11: ground-state absorption spectrum (solid line); transient depletion spectrum in region of ground-state absorption (long dashed line); T-T absorption spectrum outside region of ground-state absorption (short dashed line); T-T absorption in region of ground-state absorption computed from HadleyKeller treatment (circle); T-T absorption in region of ground-state absorption deduced from isosbestic points in the depletion spectrum (solid triangle).
.
PYS-vPy
PYS-VPY
pH 4.0
TABLE I V Extinction Coefficients at the Wavelength of Maximum Absorption for PMA-3vPy* and PMA-vPy"
20
PSS-VPy pH 11.0
80
40
Pyrens Benzene
20
polymer PMA-3~Py* PMA-vPY"
PH
11.0 11.0
X,,,/nm
418 465
a,,,/dm3
mol-' cm-l
13000 f 16OOb 10400 f l 0 W
Values are weighted means, and the errors are standard derivations of the mean. *Five concentrations were used for measurement: 1.5, 1.2,1.1, 1.0,and 0.55g/L. For py" in ref 15,,,A = 449 nm and ,e, = 60000 dm3 mo1-I cm-I. eThree concentrations were used for measurement: 1.6,1.0,and 0.8 g/L. For 3Py* in ref 9a, ,,A = 415 nm and emar = 36 400 f 14 700 dm3 mol-I cm-l. (I
860 480 580 680 760
A (nm) Figure 8. Comparison of T-T absorption for pyrene in benzene or pyrene-tagged copolymer, immediately after laser pulse.
The fluorescence spectra of our pyrene polymers are slightly sensitive to pH, similar to the results of Chu and Thomas.24" This implies that there is some change in the local environment of the pyrene as the polyacid is deprotonated, but as we will argue later, the protonated polyacids do not protect the pyrene from close encounters with quenchers. Transient Absorption. The number of polymer backbones with covalently attached pyrene groups available to us for this work was much larger than for anthracene (see Chart I). In all cases vinylpyrene was copolymerized with the appropriate majority monomer, so the differences in the spectroscopy or electron-transfer reactions described in the following must arise from modification in the local environment of the pyrene moiety. It is possible to excite pyrene using either 308-nm (XeCl) or 351-nm (XeF) excimer laser radiation, and we have carried out experiments under both excitation conditions. For equivalent excitation energies the results are sometimes dependent on excitation wavelength, which will be discussed later. It was necessary to establish the conditions for photoionization and establish the spectral shape for the 3Py* and Py'+ species. This work was carried out primarily using 308-nm radiation, for which higher energies were available. Transient spectroscopy with high laser energy carried out in the presence of traces of O2yields the spectrum of Py'+ since the triplet is rapidly quenched. Also O2removes e-(aq), which is easily observed in basic solution. By combining spectra at different time delays it is possible to extract the individual components, which are presented in Figure 7 for = 465 nm looks very PSS-vPy. The Py'+ spectrum with A, much like the major band described by Shida in low-temperature matrices15or for pyrene solubilized in water by bovine albumin,I6 poly(methacrylic acid),17or caffeine.l* From studying the relative (1 5 ) Shida, T. Electronic Absorption Spectra of Radical Ions; Elsevier Science Publishers: Amsterdam, 1988. (16) Cooper, M.; Thomas, J. K.Radiat. Res. 1977, 70, 3 12. (17) Chen, T. S.;Thomas, J. K. J. Polym. Sci., Polym. Chem. Ed. 1979, 17, 1103.
strength of the Py" and 3Py* components of the transient absorption spectrum, it was determined that an energy level of ca. 3 mJ/cm2 greatly reduced photoionization, but it was never possible to totally eliminate Pya+. The T-T spectra for all our polymer systems and pyrene in benzene are presented in Figure 8 for excitation at 35 1 nm and ca. 3.0 mJ/cm2 a t the pH indicated (although we have not observed any significant pH effect on these spectra). We found that the relative peak heights in the T-T spectrum are very similar for different polymer systems although in some cases (PSS-vPy, PAA-vPy) the ratio of these peaks was modified slightly when excited at 308 nm. The variation of the T-T spectral ratio with environment has been discussed by other authors.19 As noted above, there is always a small Py" component present at -465 nm, and we speculate that there is a single-photon route to photoionization in Py exposed to H20. Extinction Coefficient. The extinction coefficients of pyrene-tagged copolymers were determined using the GSD-HD method as described for PMA-A (see Figure 9). The extinction coefficient results for PMA-vPy copolymer are listed in Table IV. As noted above, we found that the intensity and shape of T-T absorption spectra from pyrene-tagged copolymers were quite similar to one another and showed little dependence upon pH. Hence, the PMA-vPy copolymer at pH 11 was chosen as sole representative for measurement of the extinction coefficient under the assumption that no change in extinction coefficient will arise from modifications in the local environment of the pyrene moiety. Traces of O2will preferentially quench PMA-%Py*, and it WAS assumed that any transient species other than PMA-vPy'+ generated immediately after laser pulse were quickly (less than 20 p s ) and completely converted back into ground states of copolymers in the presence of 02.If this is true, then the decay curves of optical density (after normalization) for both the ground-state and PMA-vPy+ species should be symmetric to the (18) Nosaka, Y.; Kira, A.; Imamura, M. J . Phys. Chem. 1981, 85, 1 3 5 3 . (19) For a compilation of T-T spectra, see ref 9a.
The Journal of Physical Chemistry, Vol. 96, No. 7, 1992
2898
400
Hsiao and Webber I
I
I
1
PMA-vPy DH 1 1 100
300 50
c)
0 r
x
200
n
360
0
460
560
660
760
0
. 3BO
460
560
660
760
A (nm)
Figure 11. Transient spectra of pyrene-taggedpolymer in the presence of 1 X M SPV at pH and time indicated.
100
0 360
460
560 (nm)
660
760
Figure 10. Transient spectra of PMA-vPy photoionized by 308-nm radiation at ca. 55 mJ/cm2 in the presence of 1 X M SPV and at time delay indicated.
time axis around OD = 0. We found that this was the case in our experiments. Under this assumption, the GSD-HK method can be applied to acquire the extinction coefficient of the PMAvPy'+ species, whose value is included in Table IV. Photoionization upon High Power of Laser. As was discussed above, we wish to avoid photoionization because the product of photoionization followed by capture of e-(aq) by the SPV would be identical to the reaction we wish to study; i.e. compare
-
+ 2hv Py'+ + e-(as) e-(aq) + SPV SPV'-
Py
-
(loa) (1Ob)
with
+ hu 3Py* + SPV Py
-
+
'Py*
3Py*
(1 la)
+ SPV-
(lib)
-+
PY'+
However, to our surprise, photoionization in the presence of 1 X 10-5 M SPV, which was the SPV concentration used for all pyrene experiments, did not always result in the production of SPV'-. This is illustrated in Figure 10 for PMA-vPy excited a t 308 nm. The most obvious conclusion is that e-(as) can react with the polymer backbone in some cases, since any impurities in the water or the SPV itself would be expected to be present for all experiments regardless of the polymer. This observation does not have any relevance to the triplet-state quenching experiments we report herein, but it is worth noting. An observation that is highly relevant to experiments with SPV is that this species undergoes an irreversible reaction when pyrene was excited at 308 nm. This can introduce experimental irreproducibility since the availability of SPV near the polymer depends on the total number of laser shots before stirring the solution. We did not find any evidence for a charge-transfer band between pyrene and SPV under our experimental conditions, although at high SPV concentrations complexation is observed.6b*ePerhaps the SPV absorption edge is shifted sufficiently that some degree of direct excitation does occur at 308 nm, with a subsequent decomposition reaction. An analogous observation has been made for methylviologen complexed with poly(methacry1ic acid) or poly(acry1ic acid) and excited at 308 nm.20 However, we do not observe any transient absorption or photoreaction when SPV and poly(methacry1ic acid) without covalently attached chromophore are excited at 308 nm. We do not observe any loss of SPV or experimental irreproducibility when using 35 1-nm excitation, and hence all electron-transfer results presented below will be for that wavelength. Electron- Transfer Quenching of Pyrene- Tagged Copolymers. Transient spectra obtained using 351-nm excitation at ca. 3 mJ/cm2 with 1 X M SPV always consisted of 3Py* at the earliest times, followed by a loss of triplet, and a growing-in of SPV'-. Usually the OD of py" in the 465-nm region was observed ( 2 0 ) Stramel, R. D.; Thomas, J. K. J . Chem. Soc., Faraday Trans. 2 1986, 82, 199.
to increase as will be discussed later (see Figure 11). The quantum yield of triplet formation (&) was estimated and listed in Table V using the relative actinometry method as described above. The yields of SPV' per 3Py*, Y,, are also reported in Table V, calculated by applying eq 9. Since the 3Py* was almost completely quenched, this yield approximates 6 ,. Generally speaking, both &sf and 4, depend on the copolymer backbone but only slightly on pH. The $isc values on different copolymers were quite comparable and ranged as follows in ascending order: PAA-vPy < PSS-vPy PMA-vPy < PMS-vPy. Values of & among the investigated copolymers ranged as follows: PMS-vPy < PAAvPY < PMA-VPY N PSS-VPY. The time dependence of all species, including Py'+, was monitored. For 3Py*at 418 nm there was always enhanced decay upon addition of SPV, but the spectral overlap with Py'+ and SPV'(see Figure 7) prevents the OD from approaching zero. The time-dependent optical density at 465 nm (Py") presented in Figure 12 is similar to that at 602 nm (SPV'-) for the same copolymer, except for the PMS-vPy copolymer (see later discussion). The concentration ratio [Py"] / [SPV'-] was also compared. It was our expectation that both the py'+ and SPV' species would be produced solely from the triplet state via a diffusional electron-transfer quenching process. This implies that values of [py"]/[SPv'-] would be close to unity and the OD buildup curves would be essentially identical at the appropriate wavelength. In Table V, we list two values for [Py"]/[SPV'-] which were obtained as follows: [Py'+],/[SPV'-] is a direct estimate of [py'+] assuming that any OD at 456 nm and 200 ps was contributed by Py'+ species which were generated from the quenching reactions 1l a and 11b. This is apparently not the case because the amount of py" exceeds the SPV'- produced. The OD at 465 nm contains a contribution from the triplet state and any photoionized Py'+ species. This latter contribution will decay only slightly after 200 ws. The second term in Table V, A[py'+]/[SPV'-], was calculated by subtracting the zero-time OD at 465 nm from the OD after 200 ps. This value was much closer to the expected ratio of unity except for the PMS-vPy copolymer. In the case of PMS-vPy less Py" was observed for all times, and the OD465(Py'+) decayed away more quickly than the OD,,,(SPV'-) (Figure 12). We note that when PMS-vPy was photoionized that e-(as) was formed but the Pya+concentration quickly decayed away. We believe that the Py'+ species is destabilized by the environment of this polymer. Possibly it reacts with the maleic acid moiety in a photo-Kolbe reaction.*'
-
Discussion Poly(methacrylic acid)-Anthracene. The charge separation properties of PMA-'A* are quite different from the earlier results for PMA-'A*. In the latter case efficient charge separation was observed only for low pH (ca. 3), where it was postulated that the collapsed polymer coil prevented intimate contact between the viologen and 'A*. This is consistent with the concept of "tight" and "loose" ion pairs put forward by Mataga et a1.22 In the present case low pH disfavors charge separation relative to high pH, which can be rationalized as the combined result of the spin-forbidden charge recombination process and the electrostatic (21) (a) Kraeutler, B.; Bard, A. J. J . Am. Chem. SOC.1977,99,7729. (b) Kraeutler, B.; Bard, A. J. Nouu. J . Chim. 1979, 3, 31. (c) Jaeger, C. D.; Bard, A. J. J . Phys. Chem. 1979,83, 3146. ( 2 2 ) Mataga, N.; Shioyama, H.; Kawda, Y. J . Phys. Chem. 1987,91,314.
The Journal of Physical Chemistry, Vol. 96, No. 7, 1992 2899
Electron-Transfer Quenching in Polymers
TABLE V: ~ Yield of Intersvstem Crossing and Charge Separation for Polymer-vPy _ _ ~ polymer" PH [SPVI, M $is: Ya'
PMA-VPV PMA-VP;~
PAA-vPY PAA-VPY PMS-VPY PMS-VPY PSS-vPy PSS-vPy
4.0 11.0 4.0 11.0 4.0 11.0 4.0 11.0
1.0 x 1.0 x 1.0 x 1.0 x 1.0 x 1.0 x 1.0 x
10-5 10-5 10-5 10-5 10-5 10-5 10-5 1.0 x 10-5
0.53 0.44 0.45 0.47 0.61 0.61 0.50 0.53
0.87 0.93 0.59 0.53 0.41 0.37 0.90 0.90
Yad
[Py"],/ [SPV'-]
0.78 0.73 0.65
1.61 1.67 1.78 1.84 1.22 1.20 1.84 1.87
g
0.36 g
1.17 g
A[Py"],/[SPV'-] 1.09 1.04 0.97 1.10 0.34 0.19 1.36 1.32
"The polymer concentration was prepared to make OD equal to 1.0 (2-cm path length) at 351 nm at all cases. The individual concentrations are as follows: [PMA-vPy] = 0.57 g/L, [PAA-vPy] = 1.13 g/L, [PMS-vPy] = 4.20 g/L, [PSS-vPy] = 1.32 g/L. b$iE was determined using relative actinometry method (see text). Y, was determined using eq 9 based on data from the OMA system. Y, was determined using eq 9 based on data from the CFKR system. e [Py"], was maximum concentration of Py'+ at time ca. 200 ps. fA[Py'+] was difference concentration of Py'+ between time r and time 0. #Not available
If one assumes a Poisson distribution for the probability of n quenchers in a micelle P(n) = (n)"e-(")/n! pH 11
o n
(12)
the fraction of 3A* that survive quenching is P(0) = e-(")
(13)
and the yield of ion pairs is given by Yc, = &(I - e-("))
100 200 300 400 600 PS Figure 12. Time-dependent optical density at 418 nm (3Py*)and 602 nm (SPV'-) for pyrene-tagged copolymer, in the presence of 1 X M SPV concentrations, all scaled to the same maximum. 0
100 200
300 400 600 0
repulsion of the SPV'- by the polyanionic PMA. In this context it is interesting that the pH has to be above 8.4 for this effect to be observed, even though at this pH one expects PMA to be completely i o n i ~ e d . " J ~ *We ~ ~ speculate that ionization is disfavored near the hydrophobic anthracene moiety in order to minimize exposure of the anthracene to the water. Not only is the yield of the charge separation process much higher at high pH, but also the kinetics are totally different (see Figure 5 ) . At pH 4 and 8.4 the buildup of SPV'- is essentially instantaneous, implying a "static" quenching mechanism, with only a minor growing-in of SPV'- at longer times. Likewise, the SPV induces only a minor increase in the decay rate of 3A*. This implies a kind of "micelle" model in which a viologen is either in the region of the protected 3A* or not, with little or no diffusion within the micelle or into or out of the micelle during the lifetime of 3A*. Furthermore, the presence of more than one viologen in this region does not enhance the quenching because only one viologen per 3A* is required for complete quenching to occur.24 (23) (a) Chu, D.-Y. Macromolecules 1984,17,2142. (b) Katchalsky, A,; Eisenberg, J. J . Polym. Sci. 1951, 6, 145. (c) Silberberg, A,; Eliassaf, J.; Katchalsky, A. J . Polym. Sci. 1957, 23, 259. (d) Okamoto, H.; Wada, Y. J. Polym. Sci., Parr A-2 1974, 12, 2413. (e) Kay, P. J.; Kelly, D. P.; Milgate, G. 1.; Treloar, F. E. Makromol. Chem. 1976, 177, 8 8 5 . (24) This is different than the Infelta-Gratzel-Thomas model (J.Phys. Chem. 1974, 78, 190) in which the quenching rate is taken to be proportional to the number of quenchers in a micelle. In their model it is assumed that intramicellar diffusion occurs. This point has been discussed by: Infelta, P. P. Chem. Phys. Left. 1979, 61, 88.
(14)
Since ( n ) is expected to increase with [SPV], Y, will tend to saturate as the quencher concentration increases, without any change in the decay kinetics. This simple model qualitatively agrees with our observations. The situation at high pH is very different. As the SPV concentration is increased, the rate of decay of 3A* and buildup of SPV'- both increase. This implies that diffusion of SPV into the region of the 3A* species is facile, which is consistent with the idea that the polyanion is completely extended, exposing the anthracene to the solvent. Even at pH 11 there is a residual static component, which suggests that the SPV is concentrated near the coil. This is expected on the basis of electrostatics and hydrophobic
interaction^.^^ The data presented in Figure 6 demonstrate that when the polymer concentration is increased while holding the SPV concentration constant, the fraction of 3A* quenched is decreased but the amount of SPY- produced is approximately constant. This can be rationalized by the following simple model: If we suppose that the SPV is entirely concentrated on or near the polymer, then the mole fraction of SPV moieties per repeating group (RG) on the polymer should be proportional to the ratio XSPV
[SPVI / [RGI
(15)
so long as this ratio is much less than unity. For the present case, this ratio is always less than 0.02. Obviously, Xspvdecreases with polymer concentration. In the context of this model it is reasonable to assume that the fraction of 3A* quenched is proportional to Xs,,: A[3A*]/[3A*]0 a XsPv a [RGI-'
(16)
However, so long as we do not significantly deplete the ground state
i 3 ~ * I oa
[RGI
(17)
so that A[3A*] = A[SPV'-]/&
a
[SPV]
(18)
Equations 16-18 are approximately obeyed, and we believe the qualitative features of this model are correct. A quantitative assessment of the yield and kinetic properties will require a much more detailed model and kinetic study which will be the subject of later studies.
2900 The Journal of Physical Chemistry, Vol. 96, No. 7, 1992
The yield of the charge separation products is often quite good, but for some data sets the calculated yield exceeds unity (see Table 11, last column). We believe this represents the magnitude of experimental error, although photoionization followed by electron capture by SPV could enhance the apparent Y, value. These values are also very sensitive to the assumed extinction coefficient for 3A*, which according to our ground-state bleaching work is reduced in PMA-A relative to anthracene in benzene (Table I). We presume that this diminution is the combined result of chemical attachment and solvent polarity. Pyrene-Tagged Polymers. For this group of polymers we were primarily concerned with investigating the effect of different polymer backbones on the 3Py*photophysics and electron-transfer reactions. The yield of charge-separated pairs per 3Py* is slightly higher than for PMA-A but is essentially independent of pH for a given polymer type. While there is a variation in Y,, of approximately a factor of 2 between polymer types (see Table V), we do not discern any pattern between polymer structure and Y,. It seems peculiar that the hydrophobic protection that seemed to be so important for PMA-3A* is absent here. The driving force for electron transfer from jPy* is approximately 0.2 V larger than for 3A*,2swhich could contribute to this effect. We speculate that the local polymer structure around the pyrene group is slightly modified relative to anthracene because of the molecular shape (see Summary). We recall that 'Py* behaved very differently from 'A* in our earlier work.6b,d,c The kinetics of charge separation for p~lymer-~Py* are also very different from PMA-3A*. First of all, for all cases and regardless of pH there is little or no static formation of SPV'(see Figure 12). In all cases there is an enhanced decay of )Py*, although the transient absorption of 41 8 nm is complicated by overlap with the tail of the 390-nm peak of SPV'- and Py" (see Figure 7). Thus, we conclude that in all cases the pyrene moiety is accessible to solvent and the reaction proceeds via a diffusional mechanism. The kinetics of the Py'+ species is more complicated. As mentioned with respect to the transient absorption spectra (Figure 8), at the lowest laser intensities we could use, there always seems to be a component of Py'+ (A, = 465 nm). Since the shape of the spectrum did not change for lower powers, it seems likely that pyrene can ionize by a one-photon process. However, as Lachish et al. have pointed out,26this is not always easily determined by a laser intensity study. The kinetics of Pya+at 465 nm are also quite different from those of SPV'- (monitored at 602 nm) (see Figure 12). Clearly there is a static component, followed by a growing-in as redox quenching occurs. However, the considerable overlap with the 3Py* absorption (see Figure 7) precludes a detailed analysis of this data. We also note that for PMS-Py the optical density at 465 nm decays rapidly, unlike all other cases. We speculate that the Pya+species can react with nearby maleic acid groups in a photo-Kolbe reaction2' Py"'
+ R-COO-
Py
+ R-COO'
R'
+ COZ
(19) but we can offer no persuasive argument why this reaction would not occur with the other acid-bearing polymers. +
+
Summary The present work has demonstrated that long-lived charge separation (eq la) can occur with good yield (&, eq 2) for triplet-state chromophores covalently bound to water-soluble polymers. For a zwitterionic electron acceptor, like SPV, one anticipates an enhancement of & for a polyanion (such as a deprotonated polyacid), but this effect has not proved to be important for pyrene-tagged polymers. It has been realized that the triplet excited state has the advantage of a spin-forbidden geminate recombination step (eq lb),27and we presume that the high di(25) (a) See: Techniques of Chemistry; Weinberg, N. L., Ed.; Wiley-Interscience, New York, 1975; Vol. V, Part 11. The oxidation potentials of pyrene and anthracene are 1.12 and 1.09 V, respectively. (b) Birks, J. B. Photophysics of Aromatic Molecules; Wiley-Interscience: London, 1970. The triplet-state energies for ,Py* and ,A* are 2.09 and 1.85 V, respectively. (26) Lachish, U.;Shafferman, A,; Stein, G . J. Chem. Phys. 1976,64,4205.
Hsiao and Webber electric constant of water is essential for rapid solubilization and stabilization of the ion pair. Thus,one of the important advantages of tagged copolymers is that this provides a tactic for dissolving hydrophobic molecules in water, while achieving a more rugged structure than classical micelles. One also has the advantage that there exists a vast array of polymeric structures that could be used. However, for the pyrene series there was no more than a factor of 3 difference in Y, as a function of polymer chain (ca. 0.3-0.9 in Table V). One of the observations that seems quite surprising is that the behavior of the PMA-anthracene polymer is so different from any of the pyrene polymers. The anthracene moiety appears to be well-protected from triplet-state quenching at pH C4, while earlier work has shown that it is possible to quench the excited singlet state with efficient charge separation, albeit at much higher SPV concentration.6b For low pH and either multiplicity the quenching seems to be primarily static for PMA-A.2* It seems likely to us that the anthracene, polymer coil, and viologen form a kind of aggregated system, similar to a micelle, within which very little diffusion occurs,even on the time scale of the triplet-state lifetime. It is possible that the concentration of viologen alters the precise nature of this aggregate, which would explain why singlet-state quenching occurs at a pH below 4, with ca. 10 mM SPV, while triplet quenching does not. Otherwise, one would have to postulate different distance and/or orientation dependencies for these two electron-transfer reactions. It is surprising that charge separation occurs at all at low pH, especially for the singlet state, in that electrostatic attraction between the ion pairs should enhance the back-reaction (eq lb). We have postulated before that for these PMA-based systems the degree of "hydrophobic protection" is just enough to permit ion pairs to be created and diffuse away with minimal back-reaction. We also speculate that COOH groups near the A" moiety may ionize, which reduces the electrostatic attraction. Pyrene-tagged polymers have been anomalous in that no charge separation occurred upon singlet-state quenching for high or low pH. We have previously speculated that PMA does not provide the same degree of hydrophobic protection for the pyrene moiety as for a n t h r a ~ e n e . The ~ ~ present results bear this out. Even at low pH the quenching of 3Py* tends to be primarily diffusional, implying a more open local polymer structure. For the triplet state a diffusional quenching mechanism does not preclude charge separation, as can be seen from the results for both anthraceneand pyrene-tagged polymers at high pH. Thus, we conclude that the way in which a polymer like PMA coils around a covalently bound species is influenced by the specific molecular shape and hydrophobicity of that species. Such a specificity of polymermolecular interaction was unanticipated by us when we initiated this work. At present, we do not have any clear idea what structural features would lead to this specificity. One hopes that molecular modeling could provide insights that would suggest specific polymer structures to take advantage of this effect. Our final comment is that the extinction coefficients for the triplet and cation radical states appeared to be strongly modified for our chromophores. We do not know whether this is a result of the covalent bonding to a polymer or the aqueous environment, but the effect is quite large in several cases.3o Note that if the literature values for t had been used without examination, we would have calculated yields of SPV' well in excess of unity. This should be a caution for other workers studying excited-state properties of polymer-bound chromophores, especially in very polar solvents. (27) (a) Gouterman, M.; Holten, D. Photochem. Photobiol. 1977, 25, 85. (b) Johansen, 0.;Mau, A. W.-H.; Sasse, W. H. F. Chem. Phys. Lett. 1983, 94, 107. (c) Olmsted, J.; Meyer, T. J. J . Phys. Chem. 1987, 91, 1649. (28) For the singlet-state results, see: Shand, M. A,; Rodgers, M. A. J.; Webber, S . E. Chem. Phys. Lett. 1991, 177, 1 1 . (29) In ref 6e the degree of static quenching for PMA-'Py* was estimated to be very small for concentrations of SPV below 1 mM. (30) The effect for py" may be exaggerated because the spectrum is much broader than in ref 16, which gives the spectrum in a low-temperature fluorocarbon glass.
J. Phys. Chem. 1992, 96, 2901-2903
Acknowledgment. This work has been supported by the Department of Energy, Basic Interactions (Grant DE-FG05-86ER 13629), and is gratefully acknowledged. The Center for Fast Kinetics is supported jointly by the Biomedical Research Technology Program of the Division of Research Resources of the
2901
National Institute of Health (RR008806) and the University of Texas at Austin. Registry NO. 1, 117897-81-1;2, 139277-36-4;3, 139277-37-5;4, 139277-35-3;5, 139277-38-6.
Polar Molecular Clusters Produced upon Photoinduced Electron Transfer in an Intermolecular Exciplex in Binary Solvents N. Kh. Petrov,* V. N. Borisenko, A. V. Starostin, and M. V. Alfimov Institute of Chemical Physics, The USSR Academy of Sciences, 1 1 7421 Moscow, USSR (Received: August 8, 1991; In Final Form: December 6, 1991)
A remarkably large magnetic field effect (MFE) on the fluorescence intensity of an intermolecular pyrene/N,N-dimethylaniline exciplex is observed in binary solvents containing components of considerably different relative permittivities. The values of the MFE,which depend on the composition of the binary solvent mixture, are maximal (18%, B = 30 mT) when the volume fraction of the polar component is -0.26 for a benzene/dimethyl sulfoxide liquid mixture. A simple model is proposed which assigns these features to the production of polar microdomains around each radical-ion pair. Solvent sorting then prevents ion pair dissociation.
Introduction It is well-known that cage recombination of radical-ion pairs (RIP) generated via photoinduced electron transfer in polar exciplexes can be affected by external magnetic fields, B 10 mT.I Originally created as a pure singlet state, the RIP may undergo spin evolution during its lifetime to form a triplet state via magnetic interactions, usually involving hyperfine interaction of the unpaired electrons of the RIP. A typical time required for altering the multiplicity of the RIP is 10 ns. Without applying an external magnetic field, the singlet state mixes with all three degenerate triplet sublevels S Tw,. The applied external magnetic field, which is large compared with the radical-ion hyperfine coupling, removes this degeneracy so that in the S/T evolution only one triplet sublevel takes part, S To. Thus, magnetic fields diminish the probability of intersystem crossing and, as a result, change the relative concentrations of both the singlet and triplet states of the RIP. Because of the law of spin constrvation, the multiplicity of the product formed by recombination is the same as that of their radical-ion precursor. Thus, recombination of a singlet RIP leads to states with singlet character which, in principle, may be electronically excited. Magnetic fields can strongly affect the spin evolution only of the geminate RIP. Pairs generated via random bulk recombination have a statistical ratio of singlet to triplet states (1:3). Therefore, the magnitude of the response of a RIP induced by an applied magnetic field increases when the fraction of geminate pairs grows, other conditions being the same. A convenient method for detecting magnetic field effects (MFE) on cage recombination of RIP is based on detecting exciplex fluorescence intensity. This is possible because even in rather polar solutions the transition from a RIP singlet state to a fluoresent exciplex2J is allowed, i.e. '(A-...D+J -+ I(A-D+)*
-
.+
-
-
where l(A-.-D+) is the singlet state of the RIP and I(A-D+)* is the exciplex. (1) Steiner, U.; Ulrich, T. Chem. Rev. 1989, 89, 51. (2) Petrov, N. Kh.; Shushin, A. I.; Frankevich, E. L. Chem. Phys. Lett. 1981, 82, 339. (3) Staerk, H.; Kuhnle, W.; Treichel, R.; Weller, A. Chem. Phys. Lert. 1985, 118, 19.
Recently Chowdhury et aL4 detected a MFE by exciplex fluorescence in binary solvents. The values of the MFE depend upon solvent composition, and for all mixtures investigated, the maximum of the MFE occurs at approximately the same bulk permittivity value, t, = 15-20. The largest relative value of fluorescence enhancement (-9%) caused by an external magnetic field has been detected in a liquid tetrahydrofuran (THF)/dimethylformamide (DMF) mixture. This enhancement is remarkably larger than that observed in a single solvent such as acetone4 (6%) or alcohols2 (2-3%). This comparatively large MFE, which seems quite puzzling at first, may be a consequence of the microheterogeneous structure of the binary mixture. Although this explanation has been hypothesized since the 1 9 7 0 ~ , ~ ~ nearly nothing is known about characteristic parameters of such structures. The purpose of this work is to provide new information on MFE in binary solvents and to consider the possibility of polar microdomains enhancing photoinduced electron transfer in a typical exciplex such as pyrene/N,N-dimethylaniline.
Experimental Results and Discussion To investigate the magnetic field dependence of exciplex fluorescence, a sample cell was placed in the magnetic field of a Helmholtz coil ( B = 0-30 mT). Steady-state exciplex fluorescence was excited by a xenon arc lamp in the absorption band of pyrene (Py) as previously described.* Fluorescence a t wavelengths longer than 500 nm was sent via a flexible light guide to the photocathode of a photomultiplier, from which the output was fed to a registration system. The required spectral bands were separated by suitable glass filters. The relative magnetic field effects, R , for a magnetic field, B, on exciplex fluorescence yield (pa, R ( B ) = [cp,(B)/(pa(B=O) - 11, which was measured with an accuracy of -0.2%. All experiments were carried out a t room temperature. Inert gas was bubbled through the solutions to (4) Basu, S.;Kundu, L.; Chowdhury, M. Chem. Phys. Lett. 1987,141,115. ( 5 ) Naberukhin, Yu. I.; Rogov, V. A. Usp. Khim. 1971, 40, 369 (in
Russian). (6) Kokorin, A. I.; Zamaraev, K. I. Zh. Fis. Khim. 1972, 46, 1853 (in Russian). (7) Bell, I. P.; Rcdgers, M. A. J. Chem. Phys. Lett. 1976, 44, 249. (8) Petrov, N. Kh.; Borisenko, V. N.; Starostin, A. V.;Alfimov, M. V. Izu. Akad. Nauk SSSR Ser. Khim. 1991, 1 1 , 2456 (in Russian).
0022-3654/92/2096-2901$03.00/0 0 1992 American Chemical Society