12220
J. Phys. Chem. B 2009, 113, 12220–12226
Conformation-Related Exciton Localization and Charge-Pair Formation in Polythiophenes: Ensemble and Single-Molecule Study Toshikazu Sugimoto,† Satoshi Habuchi,† Kenji Ogino,‡ and Martin Vacha*,† Department of Organic and Polymeric Materials, Tokyo Institute of Technology, Ookayama 2-12-1-S8, Meguro-ku, Tokyo 152-8552, Japan, Graduate School of Bio-Applications and System Engineering, Tokyo UniVersity of Agriculture and Technology, Koganei, Tokyo 184-8588, Japan ReceiVed: June 30, 2009; ReVised Manuscript ReceiVed: July 14, 2009
We study conformation-dependent photophysical properties of polythiophene (PT) by molecular dynamics simulations and by ensemble and single-molecule optical experiments. We use a graft copolymer consisting of a polythiophene backbone and long polystyrene branches and compare its properties with those obtained on the same polythiophene derivative without the side chains. Coarse-grain molecular dynamics simulations show that in a poor solvent, the PT without the side chains (PT-R) forms a globulelike conformation in which distances between any two conjugated segments on the chain are within the Fo¨rster radius for efficient energy transfer. In the PT with the polystyrene branches (PT-PS), the polymer main PT chain retains an extended coillike conformation, even in a poor solvent, and the calculated distances between conjugated segments favor energy transfer only between a few neighboring chromophores. The theoretical predictions are confirmed by measurements of fluorescence anisotropy and fluorescence blinking of the polymers’ single chains. High anisotropy ratios and two-state blinking in PT-R are due to localization of the exciton on a single conjugated segment. These signatures of exciton localization are absent in single chains of PT-PS. Electric-field-induced quenching measured as a function of concentration of PT dispersed in an inert matrix showed that in well-isolated chains of PT-PS, the exciton dissociation is an intrachain process and that aggregation of the PT-R chains causes an increase in quenching due to the onset of interchain interactions. Measurements of the field-induced quenching on single chains indicate that in PT-R, the exciton dissociation is a slower process that takes place only after the exciton is localized on one conjugated segment. Introduction Conjugated polymers represent a class of organic semiconductors with a wide potential for applications in optoelectronic devices, such as organic light-emitting diodes,1 solar cells,2 or field-effect transistors.3 Together with the technological development, great effort has been made to understand the photophysical properties of conjugated polymers. The study of photophysics is complicated by the fact that on a microscopic scale, each polymer chain is in a unique conformational state, and the disordered nature of the material produces a wide distribution of microscopic optical properties and interactions. Conjugated polymers contain hundreds to thousands of monomer repeat units. To account for the polymer photophysical properties, it is generally assumed that the long polymer chain is interrupted by chemical or topological defects that divide the chain into a number of sections of uneven lengths.4 The defects cause localization of the π-electrons on individual sections, creating conjugated segments that act as primary units (chromophores) in the interaction with light. The photophysical properties are determined by the number of segments, their size distribution, and by intersegment interactions.5 Primary excitation of the conjugated segment is a singlet exciton6 that decays radiatively or nonradiatively via triplet states. The excitation energy can be also transferred by dipole-dipole interactions (incoherent hopping) to other segments, either along the polymer * Corresponding author. Fax: (+)81-3-5734-2425. E-mail: vacha.m.aa@ m.titech.ac.jp. † Tokyo Institute of Technology. ‡ Tokyo University of Agriculture and Technology.
chain or between segments located on adjacent chains. The interchain energy transfer is significantly more efficient compared to the intrachain transfer7,8 due to the closer proximity of the finite-length segments. The intrachain transfer is assumed to occur only on short length scales corresponding to a few conjugated segments.9 There is, however, theoretical and experimental evidence that challenges the above picture. Although chemical modifications such as sp3 defects do interrupt the π-electron conjugation,10,11 simple bending or torsion defects do not cause significant localization of the excitation.11,12 It has been found that bent conjugated segments do exist and are characterized by shifted and broadened emission spectra.13 Onchain energy transfer much exceeding the several-segment range mentioned above has also been reported for conjugated polymers used as chemical sensors, with exciton diffusion length of more than 90 nm,14 and in the extreme case of rigid straight polymers, the wave function coherence length can reach micrometers.15 The energy transfer is also not necessarily simple dipole-dipole hopping. Coherent energy migration that can be interpreted as transfer of the conjugated segment has recently been reported.16 Singlet excitons in conjugated polymers are Frenkel-type excitons characterized generally by large binding energies on the order of 0.4-0.6 eV.17 To separate the exciton into positive and negative charge pairs requires excess energy either by excitation into hot vibrational states18 or as an externally applied electric field.19 The measured binding energy for polythiophene, for example, is 0.6 eV.20 Despite this fact, it has been reported that charge pair formation takes place from relaxed excitons, even in the absence of an external electric field.6,21,22 A model
10.1021/jp9060945 CCC: $40.75 2009 American Chemical Society Published on Web 08/19/2009
Excitons, Charge-Pairs in Polythiophenes has been proposed to explain this phenomenon.22 It is not clear, however, what is the role of the respective intrachain and interchain excitons19 or what is the contribution of energy transfer to the exciton dissociation.22 The study of carrier dynamics in conjugated polymers is another area in which different approaches lead to contradictory results. It is believed that π stacking between polymer chains is the determining factor for efficient charge transport. For example, shorter side chains attached to polythiophenes23,24 or polyfluorenes24 lead to increased charge mobilities in the solid phase due to closer main-chain packing, whereas longer substituents cause a decrease in mobility. For other systems, high mobilities along isolated single chains have been reported,25 with the charge moving freely along distances of tens of nanometers.26 An important clue to the described contradictions in exciton migration, exciton dissociation, and charge transport in conjugated polymers is the conformation of the polymer main chain and the interchain interactions. An efficient way to study the conformation-related photophysics is the use of a singlemolecule spectroscopic technique.27 The method has been widely applied in the study of various photophysical processes in conjugated polymers,28-43 including the effect of external electric field.44-47 However, only a few of these works have directly addressed the relationship between the polymer conformation and optical properties by determining the conformation theoretically or experimentally.29,30,36,42 In this report, we study two derivatives of the conjugated polymer polythiophene that enable us to control the chain conformation by the chemical composition of the polymer. The aim of the paper is to help clarify the effect that conformation of the chain has on energy transfer and exciton localization, the question of the primary pathway for exciton dissociation, and the role of energy transfer in it. One system is a graft copolymer consisting of a polythiophene backbone and polystyrene branches (abbreviated PT-PS). The presence of the branches prevents the chain from forming collapsed globulelike conformations in poor solvents. For comparison, we study the same polythiophene compound (abbreviated PT-R) taken from the synthetic process just before grafting of the polystyrene branches. We first carry on molecular dynamics (MD) simulations to determine the typical conformations of single isolated chains of the two polymers in a poor solvent. We then proceed to measure fluorescence anisotropy and fluorescence intermittency on single PT-PS and PT-R chains dispersed in an inert matrix (preliminary results of which were reported before39) to study the effect of conformation on energy transfer and exciton localization. Finally, we measure electric-field-induced fluorescence quenching of the dispersed conjugated polymers as a function of concentration, down to the level of single chains. The fluorescence quenching experiments enable us to discuss the intra- vs interchain exciton dissociation mechanisms as well as the role of energy transfer in the process. Experimental Section Molecular Dynamics Simulations. The beads-on-spring model in the coarse-grain MD method was used to simulate conformations of the two polythiophene (PT) derivatives in a poor-solvent matrix. The simulation results were further used to calculate distributions of distances between conjugated segments in the two kinds of polymer chains. The simulations were carried out using the software OCTA with the simulation engine COGNAC48 for a beads-spring model. The conditions were set so as to closely resemble the real PT-R and PT-PS
J. Phys. Chem. B, Vol. 113, No. 36, 2009 12221
Figure 1. Chemical structures of (a) PT-R and (c) PT-PS. Ensemble absorption and fluorescence spectra of (b) the PT-R and (d) PT-PS measured in dichloromethane solution (black) and in cast PMMA films (gray) at a concentration of 10-6 M.
chains. The experimental molecular weight of PT-R gives an average chain length of 95 monomer units. The length of the PS branch corresponds to 29 monomer units. The number of beads was consequently set to 95 in the main chain and to 29 in the side chains, with one side chain extending from each PT bead. The stiffness of the chain was simulated by the bending potential EB ) bθ2, with the angle θ formed by three consecutive beads set to 120° and with b ) 10 kT rad-2. The poor solvent effect on PT was simulated by the depth of an interbead Lennard-Jones potential with EPT-PT ) 0.8 kT for the PT-PT interactions. For the PS-PS interactions, the potential was set to EPS-PS ) 0.24 kT. The interbead distances were set to 0.6 nm for PT beads and 0.4 nm for the PS beads. The interaction energy EPT-PS between the PT and PS beads was obtained from the equation24 EPT-PS ) (EPT-PT · EPS-PS)1/2. Each PT-PS conformation was obtained in 20 000 simulation steps; each PT-R conformation, in 500 000 steps. Sample Preparation. The polythiophene derivatives PT-PS and PT-R were synthesized as reported before49 and characterized using optical and NMR spectroscopy. Samples for single molecule fluorescence anisotropy measurements were prepared by dispersing a small amount of the PT solution in 1.5 wt % dichloromethane solution of poly(methyl methacrylate) (PMMA) and spin-coating on cleaned quartz substrates. The resulting film was 120 nm thick. After spin-coating, the samples were vacuumdried at 60 °C for a few hours. Samples for electric-field-induced luminescence quenching were prepared by first spin-coating a 100 nm layer of poly(vinyl alcohol) (PVA) on a clean indium-tin oxide (ITO, 100 Ω/sq, 56 nm) coated glass coverslip, followed by spin-coating the PT/PMMA layer of appropriate concentration as described above, on top of which another layer of PVA (100 nm) was spin-coated, and the sample was dried in vacuum at 60 °C for a few hours. Finally, a 200 nm aluminum layer was evaporated onto the top, and the sample was immediately used for experiments. The PVA sandwich layer served to prevent charge injection during the application of electric field. Samples for bulk spectroscopic characterization were prepared by mixing the PT solution with 7.5 wt % dichloromethane solution of PMMA, casting the resulting solution into a clean Petri dish and letting the solvent slowly evaporate. The chemical structure of the compounds and their bulk absorption and emission spectra in solution and in the cast PMMA films are shown in Figure 1. Single Molecule Fluorescence Anisotropy. The fluorescence anisotropy experiments were performed using a sample-scanning confocal microscope with a circularly polarized 442 nm line of a He-Cd laser as an excitation source. The sample was placed in the focus of a microscope objective (Olympus UMPlan-
12222
J. Phys. Chem. B, Vol. 113, No. 36, 2009
Sugimoto et al.
Figure 2. Molecular dynamics simulation results. Distributions of all intersegment distances calculated from 11 simulated conformations of (a) PT-R and (c) PT-PS. Dashed vertical lines represent the estimated Fo¨rster radius. Insets: typical chain conformations. Distributions of the distances between the nth and n + ith conjugated segments along the chains of (b) PT-R and (d) PT-PS.
FL100×, N.A. 0.95) at the end of a scanning piezo tube. The fluorescence was split by a polarizing beamsplitter cube, and the two orthogonal components were detected with avalanche photodiodes (EG&G, SPCM-AQR-14). The experiments were carried out in a nitrogen atmosphere to prevent photobleaching. Electric-Field-Induced Fluorescence Quenching. Both ensemble and single-molecule experiments were carried out on a wide-field inverted fluorescence microscope with an oil-immersion objective lens (Olympus UPlanFLN100×O2, N.A. 1.3) and a circularly polarized 488 nm line of an Ar+ ion laser. The fluorescence was detected with an electron-multiplication (EM) CCD camera (Andor iXon). An electric field was applied between the ITO and aluminum electrodes in reversed bias in 5 s intervals. Results and Discussion Chain Conformation Simulations. Coarse-grain MD simulations provide an insight into the differences in chain conformations of PT-R and PT-PS in a poor-solvent matrix. Two typical examples of the simulated conformations are shown in the insets of Figure 2a, c. The PT-R molecule forms a collapsed globulelike conformation. The PT-PS molecule as a whole also resembles a globule, but the large number of the PS chains (gray) causes the main chain PT (black) to effectively stay in an extended coil-like conformation. We have confirmed that
the value of 0.8 kT of the EPT-PT is not critical and that very similar results are obtained for the EPT-PT values between 0.6 and 1.0 kT. The obtained conformations enabled us to estimate distances between individual conjugated segments that are crucial in determining the intramolecular interchain photophysical interactions. It has been found experimentally50 on a series of oligothiophenes that the length of one conjugated segment in PT corresponds, on average, to seven monomers. Accordingly, to calculate the intersegment distances, we divided the main PT chain into segments of seven beads. We neglected the distances between neighboring segments and calculated the distances along the chain between the first and third segment, the second and fourth segment, or generally the nth and n + second segment. For each segment pair, the shortest distance (given by the distance between the closest beads) was taken into account. The calculations were repeated for 11 simulated chains each for PT-R and PT-PS and are summarized in histograms in Figure 2b, d under the label n, n + 2. The process was then repeated for distances between the nth and n + third, the nth and n + fourth, and so forth, until the nth and n + 12th segments and summarized in the histograms in Figure 2b, d under the appropriate labels. All intersegment distances from the nth and n + second to the nth and n + 12th analyzed from the 11 simulated chains were then summed up and displayed in Figure 2a, c for PT-R and PT-PS, respectively. The vertical
Excitons, Charge-Pairs in Polythiophenes dashed lines in the Figure 2 indicate the estimated Fo¨rster radius; that is, the distance between two segments at which the probability of dipole-dipole energy transfer is 0.5. The radius was estimated for two randomly oriented segments using the PT-PS absorption and emission spectra and the thiophene oligomer photophysical parameters.50 The simulation results in Figure 2 a, b well illustrate that in the PT-R chains, all intersegment distances are within the Fo¨rster radius, that is, any two conjugated segments located arbitrarily on the chain are positioned for efficient pairwise Fo¨rster energy transfer. In real chains, there is a distribution of the segment lengths and of HOMO and LUMO energies, and the favorable conditions for efficient energy transfer cause fast localization of the excitation energy on the lowest-energy conjugated segment from where the energy is emitted as fluorescence. On the other hand, the results in Figure 2 c, d show that in the PT-PS chains, the efficient pairwise energy transfer occurs only up to the forth neighbor on the chain (n to n + 4). Segments further on the chain are mostly inaccessible for direct Fo¨rstertype energy transfer. As a result, in the PT-PS single chains, the exciton localization on one segment is not expected to occur, and the energy is emitted as fluorescence from multiple segments. The distributions of the intersegment distances for the PT-R chains have their maxima between 1 and 1.5 nm. However, there is a cut on the short-distance side of the distributions below 0.5 nm. This cut is due to the finite size of the beads and is an artifact due to the coarse-grain MD method used. Ensemble Spectral Properties. Chemical structures of the two polymers together with their ensemble absorption and fluorescence spectra measured in dichloromethane solutions and in cast PMMA films are shown in Figure 1. Both PT-R and PT-PS exhibit very similar absorption and fluorescence spectra in solutions, with identical fluorescence maxima at 551 nm and with absorption maxima at 419 nm for PT-R and at 426 nm for PT-PS, respectively. The similarity of the spectra indicates that in solutions, the main PT chains of both polymers are in similar extended coil-like conformations. The spectra of PT-PS change little between the solution and the cast film (Figure 2d), which confirms the simulation results that in a solid matrix, the main PT chain of PT-PS retains the extended coil conformation. The spectra of PT-R in the cast films exhibit broadening and a red shift in absorption and broadening and a blue shift in fluorescence. These observations are related to the conformational change to a collapsed globule in the films. One possible explanation is that the globule conformation results in a larger distribution of conjugated segment lengths with a greater number of shorter segments, causing a blue shift in the emission spectra. As has been shown by measuring emission spectra from single PT-R chains, the emission band of the film is inhomogeneously broadened;39 that is, there are chains that consist primarily of shorter segments and have shorter-wavelength spectra while other chains contain longer segments and are shifted toward longer wavelengths. In addition, the close chain packing in the globules can result in the formation of nonemissive aggregate states (π-π stacking) in some of the chains that work as quenchers. It is well-known that addition of quenching defects to a polymer film results in a blue shift of emission because energy transfer to the lowest energy sites is disrupted by the quenching.51 At the same time, the presence of the aggregates would cause broadening and a shift of the absorption spectra to the red. The inhomogeneous nature of the structure means that the aggregates might not form in all chains: in those chains where the aggregates do not form, the emission is not
J. Phys. Chem. B, Vol. 113, No. 36, 2009 12223
Figure 3. (a) Histograms of single-molecule emission polarization anisotropy ratio R for the PT-R (gray) and PT-PS (black) molecules. Each histogram was obtained by analyzing 140 molecules. Fluorescence intensity time traces of (b) PT-R and (c) PT-PS.
quenched by energy transfer, and an apparent blue shift in the ensemble emission spectra is observed. Single Molecule Fluorescence Anisotropy and Blinking. The phenomenon of exciton localization due to efficient energy transfer that was predicted by the MD simulation can be verified experimentally by measuring emission polarization anisotropy on single chains of the two polymers. The fluorescence anisotropy ratio, R, is defined as R ) (IP - IS)/(IP + IS), where IP and IS are the orthogonal polarization components of the emission intensity measured by the two detectors. Results obtained for 140 molecules of each polymer are summarized in a histogram in Figure 3a. The histogram of the PT-R sample has a typical U-shape characteristic of single transition dipole emitters; most molecules show large R values close to 1 or -1. This observation confirms the fact that in PT-R, the excitation energy is emitted from one conjugated segment on which the exciton localization occurs. On the other hand, the histogram of the PT-PS sample has a maximum close to zero anisotropy values; that is, most molecules emit unpolarized fluorescence. This corresponds to the theoretically predicted extended PT chain conformation in which energy transfer occurs between a limited number of segments and in which multiple randomly oriented chromophores emit simultaneously. An independent way to confirm the exciton localization is to record the time traces of fluorescence from single PT chains. Examples of such time traces for the PT-R and PT-PS molecules are shown in Figure 3b and c, respectively. In the case of PT-R, the emission intensity oscillates between a background and a maximum intensity level before a complete photobleaching takes place. This behavior is an indication of the fact that the emission of the whole PT chain proceeds from one conjugated segment that is populated by direct excitation and by energy transfer from other segments. Reversible quenching of this segment causes the observed on-off fluorescence behavior. On the other hand, emission intensity of most of the PT-PS molecules decays exponentially due to photobleaching without any apparent blinking, pointing to the existence of multiple emitters that photobleach gradually during the continuous excitation. Electric Field Induced Fluorescence Quenching I. Ensemble Experiments. The electric-field-induced quenching was carried out on a series of samples of PT-R and PT-PS dispersed at different concentrations in the PMMA matrix as a function of the applied voltage. As mentioned in the Experimental
12224
J. Phys. Chem. B, Vol. 113, No. 36, 2009
Sugimoto et al.
R ) I0R /I0′
Figure 4. (a) Electric-field-induced quenching of fluorescence for PT-R (gray) and PT-PS (black) dispersed in PMMA films at a concentration of 9 × 10-8 mol/cm3 and a field of 0.83 MV/cm. The top diagram indicates the intervals when the field was switched on. Inset: The quenching plotted on an absolute intensity scale, together with the definition of symbols. (b) Electric field intensity dependence of the quenching parameter, Q, quenching for PT-R (gray) and PT-PS (black). Sample concentration same as in part a.
Section, care was taken to electrically isolate the PT-containing matrix layer from both electrodes to prevent injection of charges. In addition, the experiments were carried out under reverse bias. We therefore use the widely accepted interpretation of fieldinduced quenching19 and ascribe it mainly to dissociation of the optically excited singlet exciton into a charge pair. A related phenomenon of luminescence quenching by energy transfer to polarons can be observed in cases that charge is deliberately injected into the system. The latter effect has been studied extensively on both the ensemble52 and single-molecule levels.45,46 In our experiments, energy transfer can partly occur to charges created by the exciton dissociation, and this possibility in PT-R and PT-PS will be discussed below. An example of the fluorescence quenching data measured on samples with a concentration of 9 × 10-8 mol/cm3 at a field of 0.83 MV/cm is shown in Figure 4a. The data are plotted in the relative (min-max) scale to highlight the differences between the PT-R and PT-PS samples. The voltage is switched on and off in 5 s intervals. At the beginning of each on interval, there is a sudden drop in the fluorescence intensity, and at the end of the interval, there is a partial recovery of the fluorescence signal. The PT-PS samples show instantaneous response to the field, whereas the PT-R samples contain also a slower component both in the quenching and in the recovery. Apart from the fieldinduced quenching, there is also gradual photobleaching of the signal. The inset of Figure 4a shows the intensity changes in absolute scale together with symbols that are used to quantify the fluorescence quenching and recovery. Using these symbols, the quenching parameter, Q, is defined as
Q ) (I0 - IV) /I0 and the recover parameter, R, as
(1)
(2)
where I0′ is the initial intensity extrapolated to take into account the effect of photobleaching. Figure 4b shows the quenching parameter measured for these samples as a function of the applied voltage. As observed also for polymers of the phenylene vinylene (PPV) family,19 the dependence is quadratic within the applied voltage range used. The quadratic dependence is characteristic of a field-induced exciton dissociation process in a completely disordered system where there is no correlation between neighboring exciton dissociation sites.19 In Figure 5a, the quenching parameter, Q, measured at the highest voltage (0.83 MV/cm) is shown as a function of the PT concentration in the PMMA matrix in the range between 2 × 10-10 and 9 × 10-7 mol/cm3. At low concentrations up to 4.5 × 10-9 mol/cm3, there is little difference between the PT-R and PT-PS samples. Since both the MD simulations and singlemolecule photophysics show that there is minimal intramolecular interchain interaction in the PT-PS, the observed field-induced quenching is due mainly to exciton dissociation that occurs effectively along the PT backbone. The limited mobility of excitons along the PT chain together with the instant response of emission intensity to the electric field indicates that the contribution to the quenching from polarons is small. The values of Q at low concentrations observed on PT-R are comparable or slightly lower than those for PT-PS. This fact can be best explained by assuming that the exciton dissociation also proceeds along the main chain (assuming otherwise would mean that the main-chain dissociation would be suppressed by a slightly less efficient mechanism). The presence of the slower response component in the emission intensity indicates that in these chains, the contribution of polaron migration to the quenching might be a more prominent (but still not the main) effect. The above explanation suggests that even in the compact globule conformation of the isolated PT-R molecules, the intramolecular interchain segment-segment interaction is not strong enough to substantially increase the field-induced quenching compared to PT-PS. This is in contrast to the exciton localization observed on single PT-R chains. Apparently, the interchain segment-segment distances in isolated PT-R chains are sufficient for long-range dipole-dipole energy transfer, but not for charge transfer, which requires shorter distances. Model calculations show that the distance between thiophene rings in a π-stacked dimer is
