AbstractWe investigate the photophysics in composite systems consisting of fullerene molecules and a conjugated polymer. Photoluminescence (PL) quenching experiments are used to study the photoinduced electron transfer that occurs after photoexcitation of the conjugated polymer. In blend systems with various fullerene concentrations we find a strong concentration dependent PL-quenching. By doping a ladder type poly(p-phenylene) (LPPP) with 5 weight % of a C₆₀-fullerene the polymer PL is quenched by more than one order of magnitude. Time-resolved measurements show that the photoinduced electron transfer can not be described by a single rate. The nonexponential PL-decay is due to a complex interplay between the diffusion of neutral excitations and their dissociation and recombination. In order to study these processes in more detail we have prepared well defined heterostructures comprising a self-assembled fullerene monolayer and a thin spin-coated polymer layer. From PL-quenching experiments on these samples we infer a value of 14 nm for the diffusion length of neutral excitations in LPPP.

Keywords: conjugated polymer, fullerene, diffusion, self-assembly, dissociation, photoluminescence, photoinduced electron transfer

1. INTRODUCTION
Composite systems consisting of fullerenes and conjugated polymers have shown promising properties for optoelectronic applications like e.g. photodetectors and solar cells ^1-3. In these materials charge carrier generation is considerably enhanced as compared to single component systems due to a photoinduced electron transfer (PET) process that can occur after photoexcitation of the conjugated polymer. After the PET the electron is located on the fullerene molecule whereas the hole remains on the conjugated polymer. Due to their spatial separation the Coulomb attraction of the oppositely charged carriers is reduced. Consequently, the generation of charge carriers is significantly enhanced as shown in recent photocurrent experiments ^{4, 5}. A more direct way to study the PET in such composite systems is the investigation of the PL properties. Since the PET strongly reduces the overlap of the electron and hole wavefunctions the probability for radiative recombination is drastically reduced leading to a significant quenching of the PL ^6-9. However, a detailed time-resolved PL study in order to unravel the dynamics of the PL-quenching is still lacking. Furthermore, most of the experiments have been limited to the investigation of simple blend systems where the fullerene and the polymer are mixed in the same solvent and then spin cast into a thin solid film. To understand the fundamental steps of the photoinduced electron transfer more defined systems than the simple mixtures are desirable. One promising approach for a controlled deposition of composite systems is the method of self-assembly ^10-12. In this technique, monolayers of a certain molecule are formed on an appropriate substrate by simply dipping the substrate into a solution containing the compound. This technique has been successfully used for the layer-by-layer growth of differently charged polyions. We have recently used this technique for the preparation of monolayers of a suitably functionalized fullerene, namely fullerene carboxylic acid (C₆₉H₈O₂) ⁹.
In our contribution we report on time-integrated and time-resolved photoluminescence measurements carried out on conjugated polymer/fullerene blend systems with gradually increasing fullerene concentrations. We show that the PL-quenching is nonexponential in nature. For a deeper understanding of the processes involved we performed experiments on a well defined model structure consisting of a self-assembled fullerene monolayer and an ultrathin spin-coated conjugated polymer layer. We found the polymer PL to be strongly quenched by the fullerene monolayer with a pronounced dependence on the thickness of the conjugated polymer film. The experimental data can be understood within a simple diffusion model. From the comparison of the experimental data with the model calculations we derive a value of approximately 14 nm for the diffusion length of neutral excitations in LPPP.

  2. EXPERIMENTAL

2.1 Sample preparation
 
 

Fig. 1: Chemical structure of the compounds used for film fabrication. The fullerene monolayer is deposited from a toluene solution containing a carboxy-functionalized C60-fullerene. A ladder-type poly(p-phenylene) conjugated polymer is deposited via spin coating. X represents a methyl-group and R and R' are sidegroups that provide high solubility of the polymer.	Fig. 2: Scheme of the LPPP/C60 heterostructures. A thin layer of LPPP is spin-coated on top of a fullerene monolayer.

 
We have used two sets of thin solid films in our experiments. Firstly, we have prepared simple blend systems by spin-casting a toluene solution containing various relative concentrations of the ladder-type poly(p-phenylene) polymer (chemical structure is shown in Fig. 1) and a standard C₆₀-fullerene. Clean glass was used as the substrate material. Secondly, we have fabricated heterostructures of a thin spin-coated LPPP film on top of a fullerene monolayer that was deposited with the self-assembly technique. Albeit there are reports that indicate that even the simple C₆₀-fullerene can be adsorbed as a monolayer on a substrate when the surface is suitably functionalized ¹⁰, a more effective deposition is obtained using a functionalized fullerene. We have utilized a carboxy-functionalized fullerene adduct, namely fullerene carboxylic acid (C₆₉H₈O₂), prepared by Diels-Alder cycloaddition ¹³. The chemical structure of this compound is shown in Fig. 1. The fullerene monolayer was deposited on a glass substrate after coating with a thin layer of poly(ethylen imine) (CH₂NH)_n (PEI), which is commercially available and often used for surface charging of substrates in self-assembly procedures. For depositing the fullerene monolayer the cleaned glass substrate was dipped into the PEI solution for 10 minutes, rinsed with pure water, spin-dried and then immersed into the fullerene solution. Thin LPPP-films with thicknesses ranging from 5 nm to 65 nm were spin-coated on top of the fullerene by varying the concentration and choosing the appropriate spin frequency. Fig. 2 shows a scheme of the samples that were used for the PL experiments. In order to directly compare the influence of the fullerene molecules on the emission of the conjugated polymer we have only covered half of the substrate with the fullerene monolayer. In this geometry, the polymer emission with and without an adjacent fullerene monolayer can be directly compared by simply moving the sample by a few millimeters.

2.2 Time-integrated and time-resolved PL-measurements

As excitation source for the PL-experiments we have used a frequency-doubled mode-locked Ti:Sapphire laser at a wavelength of 400 nm. The laser produces pump pulses of a duration shorter than 150 fs at a repetition rate of 82 MHz. Typical averaged excitation intensities on the sample were about 15 W/cm²; a value low enough to avoid nonlinear effects. The time-integrated PL-spectra were detected by a cooled CCD-spectrometer. The time-resolved measurements were carried out by the technique of time-correlated single photon counting with a microchannel plate photomultiplier tube (MCP). The temporal response of the system is limited to 50 ps. To prevent the sample from photooxidation the measurements were performed in a vacuum chamber which was maintained at a pressure below 10^-4 mbar at room temperature.

The time-integrated PL-spectra of C₆₀/LPPP blend systems with different fullerene concentrations are depicted in Fig. 3. The uppermost curve shows the PL-spectrum of the undoped LPPP. The purely electronic 0-0 transition at about 460 nm is followed by the vibronic progressions at longer wavelengths. The spectra for the C₆₀-doped samples indicate that the polymer PL is efficiently quenched for increasing fullerene concentrations. No changes in the spectrum are observed as expected for a system where all the emission bands originate from the same excited state. At a concentration as low as 0.5 % the PL is already reduced by more than a factor of two. This efficient PL-quenching is attributed to the electron transfer which occurs after photoexcitation of the conjugated polymer.
 
 

Fig. 3: PL-spectra of LPPP/C₆₀- blends with different C₆₀- concentrations. 

a)

b)

Fig.4: Time-resolved PL-measurements of LPPP/C₆₀ blends with different mixing concentrations. Fig. 4a) depicts the data on a linear scale whereas the normalized luminescence transients for three selected C₆₀-concentrations are shown in a semilogarithmic plot in the diagram 4b). 
 

To obtain a deeper understanding of the dynamics of the PL-quenching we have time-resolved the emission after pulsed photoexcitation. Fig. 4 delineates the luminescence transients for the different samples on both a linear (Fig. 4a) as well as on a semilogarithmic (normalized) scale (Fig. 4b). The linear plot shows that the PL-intensity in the presence of C₆₀ is significantly quenched already within the time resolution of the system (about 50 ps). From the uppermost curve in the semilogarithmic plot (Fig. 4b) it is obvious that the PL-decay is highly nonexponential even without any C₆₀-content. This can be explained by a trap induced nonradiative process (most likely an electron capture into an electron accepting carbonyl-group). This mechanism leads to a PL decay faster than expected for a purely radiative transition into the electronic ground state which should result in an exponential decay with a time constant of more than a nanosecond. The nonexponential nature of this process is explained by the broad distribution of distances between the trap states and the photoexcited segments of the conjugated polymer. The photoexcitations have to diffuse to the trap state prior to the nonradiative deactivation and additionally the capture process exhibits a pronounced distance dependence. Thus a broad range of rates governs the nonradiative PL-quenching processes even in the case of the conjugated polymer without any C₆₀ ^{14, 15}.
In the presence of a molecular dopant that acts as an electron acceptor an additional PL-quenching process occurs. Besides the ultrafast PL-quenching within the first 50 ps, this effect also causes an accelerated PL-decay on the nanosecond time scale as apparent from the curves for the fullerene doped polymer in the semilogarithmic plot in Fig. 4b. From these observations we have to conclude that the electron transfer is again controlled by a rather broad range of rates. This is not surprising since excitation dissociation in the presence of electron acceptors has to be described in the same framework as already discussed for the explanation of the nonexponential PL decay in the undoped samples. Again diffusion and a distance dependent capture process underlies the phenomenon. The neutral excitations perform a random walk through the inhomogeneously broadened density of states. If a C₆₀-molecule is reached within the excitation lifetime an electron transfer occurs. This process competes with the complex superposition of radiative and nonradiative processes that govern the emission dynamics of the undoped conjugated polymer. An additional problem for a quantitative interpretation of the data arises from the fact that a tendency for phase segregation is observed in such blend systems for high fullerene concentrations. Thus the density of electron acceptors and therefore also the mean distance to the next fullerene molecule might not be well defined.
In order to investigate excitation dissociation in a more quantitative manner it is crucial to have a better control of the spatial distances involved. We have therefore fabricated LPPP/C₆₀-heterostructures comprising a self-assembled fullerene monolayer below a spin-coated LPPP-film with different thickness. The absorption spectra of few selected samples with thicknesses between 5 nm and 65 nm are shown in Fig. 5. The spectra are very similar for all samples. Even for the 5 nm film no significant change in the spectrum is observed. This rules out a delocalization of the wavefunction on this length scale since no quantization effects (which should result in a blue shift of the optical transitions) are observed.
The thicknesses of the samples were measured with X-ray reflectivity. Additional atomic force microscopic studies have revealed that even the thinnest film exhibits a low surface roughness of less than 1 nm. The linear correlation between the film thickness and the optical density is shown in the inset of Fig. 5.
 

Fig. 5: Optical absorption spectra of LPPP films with various thicknesses. As an inset we show the linear dependence between the optical density at 452 nm and the film thickness.

 
 

Fig. 6: Photoluminescence spectra of a LPPP-film 35 nm thick with and without an adjacent self-assembled fullerene monolayer.

 

For the investigation of the excitation dissociation at the conjugated polymer/fullerene interface we have systematically studied the photoluminescence of the full set of samples. Fig. 6 compares the emission spectra of a LPPP-layer 35 nm thick with and without an adjacent fullerene monolayer. A strong reduction of the polymer PL due to the presence of the single fullerene monolayer is observed over the entire wavelength range. As discussed above for the blend systems, the PL-quenching is not only due to a direct photoinduced electron transfer. Excitations which are created far away from the interface have to diffuse prior to the dissociation process. The relative PL-quenching values are depicted in Fig. 7 for the different polymer films. For the 5 nm thin LPPP-film the PL is almost totally quenched, whereas the PL of the thickest film investigated here (approx. 65 nm) is only reduced by 15 %.
The experimentally found thickness dependence of the PL shown in Fig. 7 can be discussed in a simple model that accounts for the generation, recombination, diffusion and dissociation of the photoexcitations in the LPPP-layer. For the quantitative analysis we use the following continuity equation for the temporally and spatially dependent density of photoexcitations n(z,t):
 

.

(1)

In Eq. (1) z is the distance from the fullerene/conjugated polymer interface and g(z,t) describes the generation process with the femtosecond laser pulse. Here the attenuation of the laser pulse during the penetration of the LPPP-layer has to be taken into account. The second term on the right hand side of Eq. (1) gives an approximate description of the superposition of radiative and nonradiative process that lead to a PL-decay in the C₆₀-free case. We have used a value of t₀=110 ps for the 1/e PL-decay time. The two last terms in Eq. (1) model the dissociation and the diffusion of photoexcitations, respectively. The dissociation process is described with a slightly modified hopping ansatz with an exponential dependence of the rate for electron transfer between two localized states:
 

.

(2)

Since any electron in a LPPP-LUMO close to the interface can hop to a large number of accepting sites, the simple exponential dependence has to be multiplied by the z-dependent prefactor shown in Eq. (2) which results from the integration of the transfer rates from one conjugated segment of the LPPP to all fullerene molecules at the interface. In Eq. (2) a denotes the delocalization length and f₀ is a constant.
If a reasonable delocalization length a of less than 2 nm is assumed the thickness dependence of the PL-quenching shown in Fig. 7 can only be understood if diffusion of excitations to the interface is taken into account. The solid line in Fig. 7 shows the good agreement which is found when a diffusion constant of D=1.9× 10^-2cm²/s is assumed. Since this constant is related to the diffusion length via
 

,

(3)

Fig.7: The relative PL-quenching of LPPP films with various thicknesses. The filled squares and circles are the experimental quenching values at different emission wavelengths. The solid line represents the calculation based on Eq. (1) and the parameters as described in the text.

The photoluminescence measurements performed on LPPP/C₆₀ blend systems have shown that the presence of C₆₀ quenches significantly the emission in the conjugated polymer LPPP. Time resolved PL experiments have revealed the complex decay dynamics that is determined by the superposition of excitation dissociation and diffusion. In order to obtain more quantitative results, well defined LPPP/C₆₀ heterostructures consisting of a self-assembled fullerene monolayer and a spin coated LPPP film with thickness in the range of 5 nm -  65 nm were prepared. The PL-quenching at the interface in these samples can be understood in a simple model considering recombination, dissociation and diffusion of the excitations. Good agreement between experimental and calculated data was found under the assumption that the diffusion length of the photoexcitation is approximately 14 nm.

We acknowledge R. Huber and W. Stadler for technical assistance and M. Koch for useful discussions. We thank T. Salditt and M. Vogel for X-ray measurements. We are grateful to the BMBF for financial support.

1.     N.S. Sariciftci, "Role of buckminsterfullerene, C₆₀, in organic photoelectric devices," Progress in Quantum Electronics 19, pp. 131-59, 1995.
2.     G.Y.K. Pakbaz and A.J. Heeger, "Semiconducting polymer diodes: large size, low cost photodetectors with excellent visible-ultraviolet sensitivity," Appl. Phys. Lett. 64, pp. 3422-4, 1994.
3.     G. Yu et al., "Polymer photovoltaic cells: enhanced efficiencies via a network of internal donor-acceptor heterojunctions," Science 270, pp. 1789-91, 1995.
4.     C.H. Lee et al., "Sensitization of the photoconductivity of conducting polymers by C₆₀: photoinduced electron transfer," Phys. Rev. B 48, pp. 15425-33, 1993.
5.     K. Yoshino et al., "Electrical and optical properties of molecularly doped conducting polymers," Synth. Met. 78, pp. 301-12, 1996.
6.     S. Morita, A.A. Zakhidov, and K. Yoshino, "Doping effect of Buckminsterfullerene in conducting polymer: change of absorption spectrum and quenching of luminescence," Solid  State Communications 82, pp. 249-52, 1992.
7.     N.S. Sariciftci and A.J. Heeger, "Reversible, metastable, ultrafast photoinduced electron transfer from semiconducting polymers to buckminsterfullerene and in the corresponding donor/acceptor heterojunctions," Int. Journ. of Mod. Phys. B 8, pp. 237-74, 1994.
8.     X. Wei, S.V. Frolov, and Z.V. Vardeny, "Studies of photoexcitations in conducting polymers mixed with C₆₀," Synth. Met. 78, pp. 295-9, 1996.
9.  C. Kallinger et al. Self-assembly of fullerene monolayers for optoelectronic applications. in Proceedings of the International Winterschool on Electronic Properties of Novel Materials. 1997. Kirchberg, Austria.
10.     Y.M. Lvov and G. Decher, "Assembly of multilayer ordered films by alternating adsorption of oppositely charged macromolecules," Kristallografiya 39, pp. 696-716, 1994.
11.     O. Onitsuka et al., "Enhancement of light emitting diodes based on self-assembled heterostructures of poly(p-phenylene vinylene)," J.Appl. Phys. 80, pp. 4067-71, 1996.
12.     H. Hong et al., "Possible evidence for quantum-size effects in self-assembled ultrathin films containing conjugated copolymers," J.Appl. Phys. 79, pp. 3082-8, 1996.
13.     U. Jonas et al., Chem. Eur. J. 1, p. 243, 1995.
14.     U. Lemmer et al., "Time resolved luminescence study of recombination processes in electroluminescent polymers," Appl. Phys. Lett. 62, pp. 2827-9, 1993.
15.     M. Yan et al., "Defect quenching of conjugated polymer luminescence," Phys. Rev. Lett. 73, pp. 744-7, 1994.
16.     J.J.M. Halls et al., "Exciton diffusion and dissociation in a poly(p-phenylenevinylene)/C₆₀ heterojunction photovoltaic cell," Appl. Phys. Lett. 68, pp. 3120-2, 1996.
17.     K. Yoshino et al., "Marked enhancement of photoconductivity and quenching of luminescence in poly(2,5-dialkoxy-p-phenylene vinylene) upon C₆₀ doping," Jpn. J. Appl. Phys. 32, pp. L357-60, 1993.

[BACK] [MAIN] [FURTHER]