Photophysics of Cyanine Dyes - ACS Publications - American


Photophysics of Cyanine Dyes - ACS Publications - American...

19 downloads 103 Views 284KB Size

2602

J. Phys. Chem. B 1997, 101, 2602-2610

Photophysics of Cyanine Dyes: Subnanosecond Relaxation Dynamics in Monomers, Dimers, and H- and J-Aggregates in Solution R. F. Khairutdinov Institute of Chemical Physics, Academy of Sciences of Russia, ul. Kosygina 4, 117334 Moscow, Russian Federation

N. Serpone* Center for Fast Laser Spectroscopy, Department of Chemistry and Biochemistry, Concordia UniVersity, Montreal, Quebec, Canada H3G 1M8 ReceiVed: July 15, 1996; In Final Form: NoVember 11, 1996X

The photophysics of three cyanine dyes (i) 1,1′-diethyl-2,2′-cyanine iodide (pseudoisocyanine, PIC), (ii) 3,3′didodecyldithia-2,2′-carbocyanine bromide (dye 1), and (iii) 3,3′-diethyldithia-2,2′-carbocyanine iodide (dye 2) have been examined by picosecond-laser photolysis in aqueous and methanolic-aqueous media. At moderately high concentration, solutions of PIC in 5 M NaCl/water contain monomers, H-aggregates, and J-aggregates; dye 1 water/methanol solutions consist mostly of monomers and H-aggregates (dimers and higher n-mers); aqueous dye 2 solutions contain only monomers and dimers. Photolysis of H- and/or J-aggregates in PIC and dye 1 cyanine solutions leads to photobleaching of the respective aggregate absorption bands and subsequently decays by biphasic kinetics. Two mechanisms are discussed for the deactivation of excited aggregates. In the first, nonradiative decay of the excited singlet states results in considerable heating of the aggregates together with their surrounding solvent shells causing the probe laser light in a pumpprobe experiment to be strongly attenuated after excitation (see, e.g., J. Phys. Chem. 1995, 99, 11952). This heating, which subsequently leads to partial dissolution of the aggregates (deaggregation), later reformed slowly on cooling, can also arise from another mechanism. That is, relaxation of singlet excited states of H-aggregates can also occur, in part, by exciton-exciton annihilation as occurs in J-aggregates at high laser pump intensities. In this case, the longer lived excited singlet states of H-aggregates, relative to those of J-aggregates, are likely due to a less efficient (slower) exciton-exciton annihilation process in H-aggregates which would infer a weaker exciton coupling in comparison to the strong coupling known to prevail in J-aggregates. For dye monomers, singlet excited state lifetimes increase (i) with an increase in the length of the aliphatic residues attached to the dye molecule and (ii) with dimerization of the dye molecules. Dimerization restricts torsional dynamics along the dyes polymethine chain and diminishes the nonradiative deactivation channel. An important conclusion from this work is that the thermal energy stored in the aggregates in a short time after picosecond-laser excitation, and the subsequent heating of the surrounding solvent shells, leading to dissolution to smaller aggregates and monomers, provides another pathway to S1-S1 annihilation in the deactivation of excited dye aggregates. This could provide an added/alternative path for the decrease of net efficiency (i.e., decrease in charge injection efficiency) in J-aggregate sensitization of silver halide grains (see, e.g., Lanzafame et al. Chem. Phys. 1996, 210, 79) in laser-imaging technologies.

Introduction Studies of photophysical and photochemical processes of dye aggregates have attracted considerable attention because large aggregates play an important role in silver halide1,2 and nonconventional color3 photographic processes, in energy transfer in photosynthesis,4 in photodynamic therapy of cancer,5,6 and in laser technologies.7 Dye aggregates are also useful in light-harvesting arrays in artificial photosynthetic systems8-9 and are potential nonlinear optical materials.10-14 Minimal size chlorophyll aggregates and chlorophyll dimers play a crucial role in plant photosynthesis.15 Most technological applications of dye aggregates are based on the size enhancement of their optical properties caused by the excitonic nature of electronic excitations in the aggregates. Self-association (i.e., aggregation) of dyes in solution or at interfaces (e.g., solid/liquid) is a frequently encountered phe* Author to whom correspondence should be addressed at Concordia University. Fax, (514) 848-2868; e-mail, [email protected]. X Abstract published in AdVance ACS Abstracts, March 15, 1997.

S1089-5647(96)02113-X CCC: $14.00

nomenon in dye chemistry owing to strong intermolecular van der Waals attractive forces between dye molecules. Aggregation dramatically changes the photophysical properties of dyes in solutions, as attested by the characteristic changes in band shape and by the large spectral shifts relative to the absorption of dye monomers, and not least the significant deviations from Beer’s law. Cyanine dye aggregates are examined extensively since cyanines are among the best known self-associating dyes. Their aggregation in aqueous solutions16,17 and on surfaces18 often results in the appearance of a very intense, narrow, and, with respect to the monomer band, red-shifted absorption band (the J-band). H-aggregates are usually characterized by a broader blue-shifted absorption band,19 whereas dimers display a strong blue-shifted band and a variably intense red-shifted band relative to the monomer absorption band(s).20,21 An early study on the nature of J- and H-aggregates in 1,1′-diethyl-2,2′-cyanine (pseudoisocyanine, PIC) solutions noted that the respective characteristic spectral bands are not the result of different spatial © 1997 American Chemical Society

Photophysics of Cyanine Dyes

Figure 1. Cartoon depicting the various structures that may be adopted by cyanine dyes in solution: (a) brick work arrangement, (b) ladder structure, and (c) staircase arrangement; a is the angle of slippage. Large molecular slippage (∼32°) results in a hypsochromic shift of the absorption. After Harrison et al. (ref 23) and Kuhn and co-workers (ref 25).

arrangements of the cyanine molecules, but rather are due to crystal transitions in the bulk single crystal.22 Recent literature, however, suggests otherwise.23 Theoretical models have related the structure of dye aggregates to their spectral properties;24-27 regrettably, their success has been limited but to a qualitative explanation of the different spectral shifts of J- and H-aggregates bands to differences in their respective geometries. Models have also been used to estimate the sizes and structures of H- and J-aggregates; however, these estimations, mostly based on indirect evidence, have been conflicting and controversial since dye aggregation depends in a somewhat complex fashion on dye structure, solvent polarity, pH, temperature, ionic strength, and concentration of dye.23 Extensive literature (see, e.g., references in ref 23) proposes that J- and H-aggregates in solution exist as one-dimensional dye assemblies that could espouse a brick work, a ladder, or a staircase-type arrangement (Figure 1). Observation of either a red- (J-aggregate) or a blueshifted (H-aggregate) band seems to depend on the angle of slippage in the staircase architecture. Higgins and co-workers28 showed that the effective size of the excitonic state of the J-aggregate of pseudoisocyanine on a poly(vinyl sulfate) film, as determined by near-field scanning optical microscopy, is limited to e50 nm. Some anionic cyanine dyes were recently examined in aqueous solutions by Harrison and co-workers.23 Aggregation leads to formation of liquid-crystal structures (mesophases), with the actual structure depending largely on the precise molecular architecture of the dye owing to variations in electrostatic, steric and/or van der Waals short-range interactions. J-aggregates represent a liquid-crystalline phase (mesophase) consisting of thousands or more of dye monomers, depending on concentration and on the mesophase structure; the precise molecular ordering remains to be elucidated.23 Minor changes in cyanine dye monomer structure have a significant effect on the number, type, and stability of the mesophase. Judging by the vast literature and the various theoretical and experimental tools used to examine cyanine dyes (under a variety of conditions) to determine the number of monomeric units in H- and J-aggregates and the structure(s) of these aggregates, it is clear that the problem continues to present a formidable challenge. For example, the number of monomers in a J-aggregate of pseudoisocyanine appears to fall anywhere between ∼50 monomers from spectral analysis29 and 20 00050 000 monomers from integrated fluorescence yield and exciton

J. Phys. Chem. B, Vol. 101, No. 14, 1997 2603 annihilation data.30 To meet the challenge, much of the recent literature has examined the photophysics of dye aggregates adsorbed onto crystallites,1,31,32 colloids,28,33-36 or polymers.37,38 It is not the intention nor the goal of the present work to address these issues. Whatever the precise nature of, or the number of dye molecules in the structure of dye aggregates, electronic relaxation of excited aggregates is typically very fast when high photon fluences available in picosecond lasers are used to excite such species.13,33,38-40 In this regard, we41 recently reported on light attenuation by aggregates in porphyrin systems when these are excited by a picosecond-laser pulse. Laser excitation caused heating of the aggregates and led to laser light scattering because of solvent temperature inhomogeneities around the aggregates. Herein we report recent results of the picosecond-laser photolysis of three cyanine dyes and their corresponding aggregates: (i) 1,1′-diethyl-2,2′-cyanine iodide (pseudoisocyanine iodide; PIC), (ii) 3,3′-didodecyldithia-2,2′-carbocyanine bromide (dye 1), and (iii) 3,3′-diethyldithia-2,2′-carbocyanine iodide (dye 2). We selected these dyes because (a) H-aggregates can easily be obtained in solutions of dye 1 in methanol by simple addition of water, dye 2 gives dimers at sufficiently high concentrations in water, and the PIC dye is a well-known J-aggregating dye;1,2 (b) because of our interest in correlating the photophysics of cyanine dyes to their molecular architecture;40,42 (c) because of their potential photosensitizing properties of wide band gap semiconductors.43 Subnanosecond relaxation dynamics have been assessed for dimers of dye 2 and for dye aggregates of PIC and dye 1; further evidence was also obtained of the considerable heating of dye aggregates following picosecond-laser photolysis.

Experimental Section The dye 3,3′-didodecyldithia-2,2′-carbocyanine bromide was available from earlier studies.40 The dyes 3,3′-diethyldithia2,2′-carbocyanine iodide and 1,1′-diethyl-2,2′-cyanine iodide

2604 J. Phys. Chem. B, Vol. 101, No. 14, 1997 (purest grade, Aldrich), and methanol (Baker, reagent grade) were used without further treatment. The water used for the preparation of all solutions was deionized and doubly distilled from a quartz still. Solutions of dye 1 aggregates were prepared by fast mixing of 0.5 mL of a methanolic solution of the dye with 4.5 mL of water. Dye 2 aggregate solutions were prepared by dissolving the dye in water. Aggregate solutions of pseudoisocyanine iodide were prepared by injecting a small volume (0.1 mL) of the PIC dye in methanol into an aqueous 5 M NaCl solution (2.9 mL). Absorption spectra were recorded with a Shimadzu UV-265 UV-visible spectrophotometer. Transient absorption spectra were taken on a picosecond-laser system that employed a double-beam frequency-doubled (532-nm pulses focused to a 2.5-mm spot on the 2-mm quartz cell containing the samples; fwhm ∼ 30 ps; unless otherwise noted the average energy was ∼2.5 mJ per pulse which corresponds to ∼3.4 × 1016 photons cm-2 pulse-1) and passively mode-locked Nd:YAG laser, coupled to a fast detection unit that comprised a SIT (silicon intensified target) vidicon and an EG&G optical multichannel analyzer (OMA 3) coupled to an IBM 486DX computer. The probe pulse was a continuum generated from the 1064-nm fundamental by superbroadening in D3PO4 to give a probe light pulse spanning the wavelength range 400-700 nm. The absorption spectra were recorded by reading the intensities of a split probe pulse to obtain It ()1 - Iabs) and I0 light intensities for attenuation calculation. The spectra at designated delay times after excitation at 532 nm are the averages of about 6-10 pairs of laser pulses. The solutions were stirred after each pulse to probe a different sampling of the solution. Other descriptive details of the picosecond-laser system are found elsewhere.44-47 Time-resolved emission decay data were obtained by the streak camera method48 using 532-nm pulsed laser excitation (fwhm ∼ 30 ps, average energy per pulse ∼ 2.5 mJ) and a Hamamatsu C979 streak camera; samples were prepared to yield optical densities of about 0.1 at this wavelength. A laser beam splitter diverted 10% of the pulse to an energy meter (Laser Precision Corp.), while the remaining light was smoothly focused onto a 4-mm2 area of the 2-mm quartz cuvette. The fluorescence was collected from the front surface and collimated by a 50-mm diameter lens (f/1.5). A second lens (focal length, 60 mm; f/1.2) focused the collected light through a 100-mm slit of the streak camera. A red-sensitive multialkali photocathode was used for detection. The streak camera was protected with a suitable cutoff filter to prevent detection of scattered and Raman-shifted pump radiation. All experiments were carried out at ambient temperature with freshly prepared, air-equilibrated solutions unless otherwise noted. Preparations of solutions of aggregates under otherwise identical conditions gave reproducible absorption spectral features. Results and Discussion J- and H-Aggregates of the PIC Dye. Pseudoisocyanine produces J-aggregates in concentrated aqueous salt solutions. Absorption spectra of 1.5 × 10-4 M PIC in 5 M NaCl/H2O (2-mm cell) at different temperatures are illustrated in Figure 2 (solid lines). The dashed line depicts the absorption spectrum of a 2.2 × 10-6 M PIC solution. At this low concentration, the absorption spectrum is characteristic of PIC monomers.49 An intense narrow J-band centered at 572 nm and a less intense broad absorption displaying two maxima (at 495 and 532 nm) are characteristic of absorption spectra of the pseudoisocyanine dye at high concentrations and at moderate temperatures. These

Khairutdinov and Serpone

Figure 2. Absorption spectra of pseudoisocyanine iodide (PIC) 1.5 × 10-4 M in 5 M NaCl/H2O (2-mm cell) at different temperatures. The dashed line is the absorption spectrum of 2.2 × 10-6 M PIC; other conditions are the same.

broad, short-wavelength absorption bands are significantly shifted in comparison to the bands displayed by solutions of PIC monomers and dimers.49 They correspond to absorption bands of H-aggregates of the dye. Heating a solution of the PIC dye in 5 M NaCl/H2O media results in a dramatic decrease of the J-absorption band and in a blue-shift of both shortwavelength absorption bands. At 55 °C, the J-band is barely discernible and the positions of the two spectral features at 495 and 532 nm converge to the two absorption maxima of PIC monomers. The implications of the temperature-dependent conversions between the monomers and the different states of aggregation of the pseudoisocyanine dye will be treated elsewhere.50 Of importance in this paper is that solution heating leads to a reversible decrease of the intensities of the absorption bands of H- and J-aggregates and to a corresponding increase of the monomer bands (can be cycled several times); the results infer the exothermic equilibria noted in eq 1.

monomers a H-aggregates a J-aggregates

(1)

Picosecond-laser excitation of a concentrated solution of PIC (1.5-4.0 × 10-4 M) in 5 M NaCl/H2O leads to bleaching of the J- and H-aggregate absorption bands, subsequently followed by their absorption recovery (Figure 3a-c), albeit partially in our time window of 20 ps to 12 ns. A short-lived, pulse-widthlimited (τ e 30 ps) fluorescence at λ > 550 nm was also detected by time-resolved techniques; the fluorescence yield (Φ scales with τ) is consistent with a S1-S1 exciton annihilation process (see below). The recovery of the solution absorption is totally reversible since no changes in absorption were detected even after 100 laser pulses. Absorption recovery of the J-band at 572 nm compares with the transient absorption decay at ∼585 nm; both follow biphasic kinetics (see Figure 3b and ref 13). The first phase consists of a very fast, pulse-width-limited partial recovery of ground state absorption, followed by a slower nanosecond recovery. Estimated times of the slower recovery and the corresponding energy dependence are illustrated in Figure 3b. Even though the changes in absorption recovery from 50 ps to 12 ns represent only ∼10-20% of the total recovery, it is nevertheless significant to note that the time of recovery decreases with a decrease in laser pulse energy.

Photophysics of Cyanine Dyes

J. Phys. Chem. B, Vol. 101, No. 14, 1997 2605 during the release of excitation energy, we estimate the temperature in cyanine aggregates to rise from ambient to ∼140 °C. This temperature would be higher if several quanta were absorbed by each J-aggregate during the laser pulse, not unlikely under the picosecond-laser intensities used (∼3 × 1016 photons cm-2 pulse-1). In the present instance, the immediate consequence of this temperature rise is a significant heating of the solvent at the interface with each aggregate. Figure 2 shows that such heating causes a decrease of the intensity of the J-band because of partial dissolution54 of the J-aggregates into smaller23 H-aggregates and ultimately into monomers (deaggregation, see eq 1). Subsequent dissipation of the thermal energy from the interface to the bulk solution decreases the temperature at the interface and leads to the recovery of J-aggregate absorption (Figure 2). The characteristic times of temperature changes at a distance a from the pulse-heated center can be estimated from eq 255

τ≈

Figure 3. (a) Transient spectra of a solution of 4 × 10-4 M PIC in 5 M NaCl/H2O (2-mm cell) at different delay times after the laser pulse. (b) Kinetics of the absorption recovery of the PIC J-band at λ ) 572 nm after picosecond-laser excitation at different energies, E, of the laser pulse. τ is the characteristic time of the slow component of the transient bleach decay curves. [PIC] ) 1.5 × 10-4 M in 5 M NaCl/H2O solution. (c) Kinetics of the transient bleach decay of a 4 × 10-4 M PIC in 5 M NaCl/H2O solution at λ ) 495 nm; τ ) 155 ( 16 ps is the characteristic time of the fast component of the absorption recovery curve.

Electronic relaxation of excited states in short time by a nonradiative channel releases a significant amount of energy in each J-aggregate, existing in a one-dimensional brick work arrangement as suggested by Kuhn and co-workers25 (Figure 1). Fast electronic relaxation noted earlier41 in porphyrinic aggregates, examined under otherwise identical photolytic conditions to those used here, resulted in a considerable discharge of the nonradiative energy to the surrounding solvation shells causing the laser light to be strongly attenuated because of the temperature inhomogeneities generated at the aggregate/ solution interface. In the present instance, relaxation of the J-aggregates excited state is pulse-width-limited (e30 ps) and arises from an efficient exciton-exciton annihilation (see below) in each aggregate.30,51-53 If all the light absorbed resulted in heating the cyanine aggregates predominantly by a nonradiative channel and if no energy were dissipated outside the aggregates

a2Fc k

(2)

where F is the density of the medium (ca. 1 g cm-3), c is the heat capacity per unit mass of the solution medium (∼1 cal g-1 °C-1), and k is the thermal conductivity ()1.4 × 10-3 cal s-1 cm-1 °C-1). At a distance a ≈ 102 nm, temperature relaxation occurs in ca. 100 ns, in accord with the estimated times of the slower J-aggregate absorption recovery (Figure 3b). The distance of 102 nm is typical of the size of local temperature inhomogeneities in the solution medium that can lead to efficient light attenuation through scattering. A decrease in laser pulse energy provides less heating of the J-aggregates and consequently to less dissolution of the Jaggregates and to a decreased fraction of the slower absorption recovery (Figure 3b). Transient absorption spectra of H-aggregates of pseudoisocyanine are also presented in Figure 3a. Absorption recovery of the PIC H-aggregates band at 495 nm also follows biphasic kinetics (Figure 3c). The first phase is characterized by a fast, albeit partial recovery of the absorption (64% at ∼0.5 ns) with a characteristic time of τ ) 155 ( 16 ps, followed by a slower recovery of the remaining (36%) ground state absorption in nanosecond time (pulse energy ∼2.5 mJ). Qualitatively, these changes are in keeping with those observed for J-aggregates and result from heating H-aggregates and their surrounding solution medium on picosecond-laser excitation generating local temperature inhomogeneities. Spectral results in Figure 2 indicate that heating decreases the intensity of the H-aggregate absorption band. It is difficult to attribute the low excited state decay time of PIC H-aggregates to any single factor. However, it is not unlikely that, at the high laser intensities used, exciton-exciton annihilation as occurs in J-aggregates30,45-47 may, in part, be responsible for the small decay times in H-aggregates. The larger characteristic time of the fast absorption recovery in the latter (see above), compared to the corresponding pulse-widthlimited recovery time in J-aggregate absorption, would be sensible if exciton-exciton annihilation were slower in Haggregates than in J-aggregates. This difference is reasonable because of the strong coupling of excitons known to prevail in J-aggregates and because of an expected weaker exciton coupling in H-aggregates. Dye 1 Aggregates. Absorption spectra of dye 1 in methanol and in water/methanol (9/1) mixtures are presented in Figure 4. The solid lines represent the normalized absorption spectra at different dye concentrations at 20 °C; the dashed line depicts

2606 J. Phys. Chem. B, Vol. 101, No. 14, 1997

Khairutdinov and Serpone

Figure 4. Normalized absorption spectra of dye 1 in methanol and water/methanol (9/1) solutions at 20 °C (solid lines). The dashed line depicts the absorption spectrum of the 3 × 10-5 M solution of dye 1 also in water/methanol (9/1) but at 60 °C.

the absorption spectrum of a 3 × 10-5 M solution of the dye also in water/methanol (9/1) at 60 °C. The absorption spectrum consisting of a band at 555 nm and a less intense shoulder on the short-wavelength side for a 3 × 10-6 M methanolic solution of dye 1 is typical of the spectrum of dye 1 monomers (Figure 4).40 In addition to the monomer band in the spectrum of moderately concentrated (3 × 10-5 M) dye 1 water/methanol solutions, there also exists a more intense band centered at about 520 nm. The large broadening of the 555-nm band, in comparison to the spectral features of the monomers, suggests the existence of an additional red-shifted absorption band clearly seen at ∼575 nm in the more concentrated (1.5 × 10-4 M) solution of dye 1 in water/methanol. The appearance of two bands, one, a more intense blue-shifted band, and the other, a less intense red-shifted band compared to the absorption features of the monomers, is characteristic of dye dimers.20,21 It is tempting, therefore, to assign the 520-nm band to dye 1 dimers in the 3 × 10-5 M solutions of the dye. However, the existence of H-aggregates with a higher number of monomers in each aggregate (i.e., n-mers) in this aqueous methanolic solution of dye 1 is not precluded by the data. In the discussion below, H-aggregate will denote dimers and higher order n-mers in dye 1 solutions. Heating the solution to 60 °C leads to the reversible decrease of the number of H-aggregates in solution and to a subsequent increase of the number of monomers (compare the solid line and the dashed line absorption spectra of 3 × 10-5 M solutions of dye 1 in aqueous methanol; Figure 4). At higher concentrations (1.5 × 10-4 M dye 1 in H2O/MeOH), a significant quantity of larger size H-aggregates (n-mers) are produced that are responsible for the very broad absorption feature at 525 nm. Time-resolved transient spectra of a moderately concentrated solution of dye 1 (3 × 10-5 M; water/methanol) are presented in Figure 5a for delay times 0-600 ps. Picosecond-laserinduced bleaching of the H-aggregate absorption at ∼525 nm is followed by its recovery via biphasic kinetics; they are summarized in Figure 5b. In the first phase, the absorption is partially recovered (about 83% at 1 ns) with τ ) 270 ( 30 ps; the remaining ∼17% absorption recovers more slowly at times longer than 10 ns. No changes in absorption were detected even after ∼100 laser pulses. The spectra of Figure 5a also display a new transient absorption feature at ∼555 nm formed in the subnanosecond time domain; this feature coincides with the band of dye 1 monomers (Figure 4). This, together with the existence of the slow recovered component of the absorption band around 525 nm, is understandable if electronic relaxation of excited H-

Figure 5. (a) Transient spectra of a 3 × 10-5 M solution of dye 1 in water/methanol (9/1) at different delay times after the laser pulse. (b) Kinetics of the absorption recovery at 525 nm; τ ) 270 ( 30 ps is the characteristic time of the fast component. (c) Transient spectra of dye 1 monomers in methanol; inset shows transient bleach decay kinetics at 560 nm; τ ) 215 ( 20 ps.

aggregates resulted in their partial conversion to monomers, owing54 to the heat dissipated to the solution medium surrounding the H-aggregates. In accordance with Figure 4, such heating is responsible for the reversible decrease of the number of H-aggregates in solution and to the subsequent increase in the number of monomers that display the spectral feature at 555 nm. The absorption recovery time at this wavelength is τ ) 190 ( 37 ps (Figure 5b). It is interesting that the decay time of the singlet excited state of dye 1 H-aggregates (τ ) 270 ( 30 ps) is similar, albeit fortuitously, to the corresponding decay time of dye 1 monomers in dichloromethane reported earlier by Serpone and Sahyun (τ ) 241 ( 55 ps).40 In the latter case, it was noted that intersystem crossing to an isomerized excited state and internal conversion played a major role in the nonradiative deactivation

Photophysics of Cyanine Dyes

Figure 6. Transient spectra of a 1.5 × 10-4 M solution of dye 1 in water/methanol (9/1). The delay time after the laser pulse is shown on each spectrum.

of excited singlet dyes, torsional motion about the polymethine chain being prerequisite to both processes.40 The lifetime of the singlet state of dye 1 monomers in water/methanol solution is somewhat smaller (τ ) 190 ( 37 ps) and identical to the transient bleach decay of dye monomer absorption in methanol (Figure 5c). The slightly faster relaxation in a more polar medium is as expected since efficient coupling of a charged excited state with a polar medium such as water enhances the nonradiative pathway (heat dissipation, see above); evidently, this nonradiative decay path is not as efficient in dichloromethane. The slower decay of the excited singlet state of H-aggregates of dye 1 in water/methanol is consistent with the notion that aggregation has diminished the torsional motion about the polymethine chain and thereby has suppressed, to some extent, one of the deactivation channels. Similar changes in the transient spectra, as depicted in Figure 5a, are also observed after picosecond-laser excitation of a more concentrated solution of this thiacarbocyanine dye 1 (1.5 × 10-4 M in water/methanol; Figure 6). In addition, considerable light attenuation is observed after the laser pulse. The efficiency of light attenuation first increases, reaches a maximum at ∼2 ns delay, and subsequently decreases on the nanosecond time scale. In accordance with our earlier observations41 on porphyrin aggregates and inferred for the pseudoisocyanine dye (see above), we deduce that the sizable picosecond-laser light attenuation seen in cyanine aggregates after excitation is also due to heat dissipation from the larger n-mer species in H-aggregates to their immediate solvent medium at the interface. To ascertain the veracity of light scattering by an irradiated sample of the more concentrated solution of dye 1 in water/ methanol, we conducted laser pulse experiments similar to those performed earlier41 using a modified scheme of the probe light path. A light-collecting lens was placed in the path immediately after the sample such that the sample was at the lens focal point. This lens had no effect on the intensity of the probe light transmitted through the sample and entering the detection system for non-light-scattering samples but did increase the intensity of the light entering the detection system for light-scattering samples. The solution was subsequently illuminated with the laser pulse, and the probe light attenuated by the cyanine sample was compared in the presence and absence of the lens (note that no such probe light attenuation takes place in the absence of the pump pulse excitation). The results indicated a significant increase in the vidicon (detector) signal of the picosecond-laser illuminated solutions of dye 1 H-aggregates in the long time domain (1-10 ns after the pulse). They clearly demonstrate that light scattering is responsible (at least in part) for the observed laser light

J. Phys. Chem. B, Vol. 101, No. 14, 1997 2607

Figure 7. Absorption spectra of dye 2 in water at two different concentrations.

extinction by the excited dye 1 aggregates within our nanosecond time window of less than 12 ns. The appearance of the new feature around 560 nm in the transient (extinction) spectra of Figure 6 (dye 1 concentration, 1.5 × 10-4 M) at delay times g200 ps is attributed to an increase in the number of dye 1 monomers because of the energy dissipated at the H-aggregate/solution medium interface and the consequent partial dissolution (deaggregation) of the larger n-mer aggregates. The existence and the ensuing heating of H-aggregates in picosecond-laser experiments are also not precluded at the more moderate concentrations of dye 1 (3 × 10-5 M; Figure 5a). Dye 2 Aggregates. Absorption spectra of dye 2 at two different concentrations (1 × 10-6 and 6.5 × 10-5 M) in water are presented in Figure 7. In the more dilute solution, the absorption spectrum is characteristic of the absorption of monomeric dye molecules. The new blue-shifted band centered at ∼510 nm that appears at the higher concentration is that of dimers.1 Figure 8a illustrates the temporal changes in the transient spectra for a 4 × 10-5 M aqueous solution of dye 2. The corresponding ground state absorption recovery at different wavelengths is illustrated in Figure 8b. All the absorption recovery curves and the transient absorption decay curve have a fast component with characteristic times in the range 100300 ps and a slower component with characteristic times much greater than 10 ns. Full recovery of dye 2 absorption at long times was confirmed by the stability of the solution even after 100 laser pulses. The upper curve in Figure 8b with a decay time of the fast component τ ) 157 ( 15 ps (error denotes 1 standard deviation) corrresponds to the decrease of the transient absorption at 435 nm at which thiacarbocyanines typically display singlet excited state absorption.40 The lowest two curves correspond to the partial recovery of the monomer absorption at ∼555 nm in water (τ ) 126 ( 8 ps) and in methanol (τ ) 120 ( 10 ps) from excited singlet states. The transient bleach decay curve with τ ) 235 ( 10 ps (510 nm) corresponds to the absorption recovery from the excited singlet of dye 2 dimers (we found no evidence for H-aggregation with higher n-mers in dye 2 solutions). Note that in contrast to dye 1, the characteristic time of absorption recovery for dye 2 monomers is half the time(s) needed to recover the absorption for the dye 2 dimers. An increase in excited state lifetime in dimers, in comparison with monomers, is often observed in dye photophysics. This is due to a decrease of the excited state nonradiative decay channel originating from the cooling/freezing of some of the vibrational and torsional modes in the dimer molecules. Indeed, excited state lifetimes of cyanine dyes are very sensitive to the rigidity of the local environment,56,57 which

2608 J. Phys. Chem. B, Vol. 101, No. 14, 1997

Figure 8. (a) Transient spectra of a 4 × 10-5 M solution of dye 2 in water at different delay times after the laser pulse. (b) Kinetics of transient absorption and transient bleach decay at different wavelengths and in different media; τ’s are the characteristic times of the fast components. The upper curve denotes the decay of the singlet excited state absorption of dye 2 monomers.

can significantly restrict the torsional dynamics around the polymethine chain. Inhibition of this torsional motion decreases the nonradiative rate constant and increases the lifetime. The longer lived transient absorption recoveries for dye 2 monomers in both water and methanol solutions occur subsequently to efficient intersystem crossing40 from the singlet excited state to the longer lived triplet state of dye 2 monomers; Φisc ≈ 0.4 in both media. In the case of excited dye 2 dimers, the longer transient absorpion recovery may arise either from the longer lived triplet excited states subsequent to intersystem crossing from the excited singlet states and/or from partial conversion54 of dimers to monomers. Finally, it is noteworthy that the excited singlet state lifetime of dye 2 monomers is smaller (by nearly a factor of 2) than the corresponding lifetime of dye 1 monomers, even though they possess identical structures of their light-absorbing chromophores. This difference is attributed to the different nature of the aliphatic residues attached to the nitrogen atoms in these otherwise identical molecules. Dye 1 differs from dye 2 only by the presence of two longer aliphatic chains in dye 1 (dodecyl

Khairutdinov and Serpone vs ethyl). The longer tails in dye 1 restrict any torsional dynamics of parts of the molecule around the polymethine chain, an important deactivation channel.40 Nature of Excited States in Aggregates and Exciton Annihilation. Few studies of dye aggregates have addressed the question of the nature of excited states involved in dye aggregates photophysics. Recently, Horng and Quitevis38 proposed a three-level kinetic model for the relaxation dynamics in polymer-bound J-aggregates of pseudoisocyanine chloride on the basis of fluorescence and transient spectra pump-probe measurements. The model involves the ground state, denselyspaced singlet exciton states, and a long-lived “bottleneck state”. The nature of this latter state has not been without controversy. Singlet and triplet states are commonly used to rationalize the photophysics (lifetimes, intersystem crossing efficiency, radiative and nonradiative rates, among others) of dye monomers in which the triplet state could be viewed as the bottleneck state in the overall excited monomer relaxation events. By analogy, we infer that singlet and triplet excitonic states can also be implicated in excited J- and H-aggregates deactivation, with the triplet states providing the same bottleneck in the relaxation dynamics. In keeping with this view, Gradl and co-workers58 observed phosphorescence enhancement of PIC J-aggregates at low temperatures consistent with the notion that aggregation enhances intersystem crossing, producing higher triplet yields and consequently higher emission quantum yields from triplet excitonic states, over those prevailing in monomers.38 To the extent that only a short-lived (e30 ps) fluorescence is observed at wavelengths greater than 550 nm from aggregates in PIC aqueous/salt solutions (see above), the bottleneck state cannot be of singlet character, a point also made earlier38 for polymer-bound PIC J-aggregates. It is useful to describe these J- and H-dye nanostructures in semiconductor terminology: the ground state is the valence band (HOMOs) and the densely spaced38 singlet excitonic states are the levels (LUMOs) of the conduction band. The bottleneck state would then be an intraband state. Low-temperature holeburning and photon-echo studies of PIC J-aggregates led DeBoer and co-workers59 to suggest the bottleneck state to be an ionized dye•+/trapped electron state; in this terminology, dye•+ would be a trapped hole. Presently available data do not permit a distinction between a molecular versus a semiconductor-type description of the relaxation dynamics of excited dye aggregate nanostructures. Further consideration should be given to the latter description. Various bimolecular processes become increasingly important for the disposal of singlet excitation energy with increasing picosecond-laser excitation intensities. Singlet-singlet exciton annihilation gives rise to various product states (eq 3)60,61

S1 + S1 f S0 + S1 + (heat)

(3a)

S 0 + T1 S1 + S1 f S0 + S0 + (heat) T1 + T1

(3b)

[ ]

where S0 is the ground state, S1 is the first excited singlet state, and T1 is the lowest energy triplet state (the bottleneck state). In each case heat is produced62 in the annihilation process to heat the aggregate and subsequently the aggregate/solution interface. Typically, resonance transfer of excitation energy can take place between neighboring dye molecules when these are separated by a distance less than ∼50-100 Å, a condition met in J- and H-aggregates. At the picosecond-laser intensities used in our experiments (0.8-3.3 × 1016 photons cm-2 pulse-1), the

Photophysics of Cyanine Dyes photon fluence and the exciton density are sufficiently high that two singlet excitons can encounter and annihilate one another during their lifetime (