Polarization Imaging of Emissive Charge Transfer States in Polymer

Nov 11, 2014 - Daniela Täuber , Alexander Dobrovolsky , Rafael Camacho , and Ivan G. ... Rafael Camacho , Sumera Tubasum , June Southall , Richard J...
0 downloads 0 Views 4MB Size
Article pubs.acs.org/cm

Polarization Imaging of Emissive Charge Transfer States in Polymer/ Fullerene Blends Rafael Camacho,† Matthias Meyer,† Koen Vandewal,‡ Zheng Tang,§ Olle Inganas̈ ,§ and Ivan G. Scheblykin*,† †

Chemical Physics, Lund University, Box 124, 22100 Lund, Sweden Institut für Angewandte Photophysik, Technische Universität Dresden, 01062 Dresden, Germany § Biomolecular and Organic Electronics, IFM, Linköping University, SE-581 83 Linköping, Sweden ‡

S Supporting Information *

ABSTRACT: Photoexcitation of conjugated polymer−fullerene blends results in population of a local charge transfer (CT) state at the interface between the two materials. The competition between recombination and dissociation of this interfacial state limits the generation of fully separated free charges. Therefore, a detailed understanding of the CT states is critical for building a comprehensive picture of the organic solar cells operation. We applied a new fluorescence microscopy method called twodimensional polarization imaging to gain insight into the orientation of the transition dipole moments of the CT states, and the associated excitation energy transfer processes in TQ1:PCBM blend films. The polymer phase was oriented mechanically to relate the polymer dipole moment orientation to that of the CT states. CT state formation was observed to be much faster than energy transfer in the polymer phase. However, after being formed an emissive CT state does not exchange excitation energy with other CT states, suggesting that they are spatially and/or energetically isolated. We found that the quantum yield of the CT emission is smaller for CT states spatially located in the highly oriented polymer domains, which is interpreted as the result of enhanced CT state dissociation in highly ordered structures.



INTRODUCTION Plastic solar cells based on blends of conjugated polymers and fullerene-based electron acceptors have attracted attention as an alternative to silicon based devices due to their potential for versatile and low-cost manufacturing.1−6 Following excitation of the polymer (donor) or fullerene (acceptor), fast electron/ hole transfer1 at the donor:acceptor interface results in a local charge separation. A debate on the role of direct charge generation through delocalized charge transfer states in the acceptor domain versus charge generation through a bound charge transfer (CT) state at the donor/acceptor interface is ongoing.7−9 The generation of fully free charge carriers by any CT state is determined by the competition between recombination and dissociation of this state. In a solar cell, following charge separation, free charge carriers travel to the electrodes, or encounter each other, again forming a CT state which can subsequently dissociate once more or recombine. Recombination of the CT state can result in emission with a radiative rate proportional to the square of its transition dipole moment. This is because of the coupling between the CT ground and excited states.10 CT emission has been detected and spectrally resolved by photoluminescence (PL)11−22 and electroluminescence (EL) experiments.14,16,17,19,21,22 CT emission results in a spectral band, generally red-shifted in comparison to the donor fluorescence, which is not present © XXXX American Chemical Society

in the emission spectra of neat samples of both donor and acceptor. However, CT emission is generally weak and thus can be masked by the residual emission from the donor and/or acceptor materials. CT states play a major role in the operation of polymer:fullerene solar cells.11,14,18,22,23 Therefore, a crucial step for a unified description of the processes that determine the power conversion efficiency of organic solar cells is a detailed understanding of the electronic structure of CT states, as well as their population and depopulation pathways.2,21,24−27 Recently, there have been efforts to study the composition of the CT state transition dipole moment through fluorescence anisotropy measurements.21 However, these studies have so far not accounted for the effects of the excitation energy transfer (EET) on the fluorescence anisotropy. For example, EET is known to reduce the fluorescence anisotropy of neat conjugated polymer films28 obscuring its connection to the transition dipole moment orientations. Therefore, we study the CT state emission by using a two-dimensional polarization imaging (2D-POLIM) technique.29,30 This technique allows us to quantitatively estimate the extent of the EET processes Received: July 9, 2014 Revised: November 10, 2014

A

dx.doi.org/10.1021/cm502503f | Chem. Mater. XXXX, XXX, XXX−XXX

Chemistry of Materials

Article

occurring in polymer:fullerene blend films, and to take it into account for unmasking the emissive CT state dipole moment composition. In this work we report the study of the polymer:fullerene poly[2,3-bis(3-octyloxyphenyl)-quinoxaline-5,8-diyl-alt-thiophene-2,5-diyl] (TQ1):phenyl-C61-butyric acid methyl ester (PCBM) blend films. Two types of films were used: isotropically oriented films, and films in which a predominant dipole orientation of the polymer material was induced mechanically. In order to obtain a large degree of alignment we were limited in our choice of stoichiometry in donor:acceptor blends (excess of polymer), and therefore cannot use the blend ratios used in optimized solar cells. We measured the polarization properties of both the absorbing polymer states and the emissive CT states, and quantified the EET efficiencies through the so-called single funnel approximation.30 This sheds light onto the composition of the emissive CT state’s dipole moment, and provides insight into the energy redistributions occurring among the polymer phase, among CT states, and between the two species. We found that the formation of the CT states is faster than the intra-EET of the donor phase. Further, the more oriented domains of the donor phase are associated with less emissive CT states. This indicates that the local order of the polymer phase enhances charge separation or results in a reduced density of emissive CT states. Simulations were used to gain insight into the EET occurring between emissive CT states and to calculate the composition of their transition dipole moment. We found that the emissive CT states do not exchange energy with other states and that they borrow more dipole moment from the donor material than previously thought.



aperture = 0.6) was used to image the sample. The samples were kept under nitrogen to avoid photoinduced oxidation. The collected fluorescence leaves the microscope through a special port, which used a mirror instead of a prism to avoid depolarization effects in emission. Two sets of emission filters were used to select the spectral range of the fluorescence to be detected. The filter set polymer allowed wavelengths between 675−810 nm while the filter set red allowed wavelengths larger than 810 nm to reach the detector. The emission analyzer consisted of a wire-grid linear polarizer (Edmund optics) mounted in a motorized rotating mount. Finally, the fluorescence emission was imaged on an EMCCD camera (Cascade 512 B). Motors and camera were controlled using LabVIEW (2012).



RESULTS AND DISCUSSION Absorption and Fluorescence Spectra. To establish the presence of CT emission in blends of the conjugated polymer TQ1 mixed with the fullerene derivative PCBM (4:1 polymer:fullerene weight ratio), we measured the absorption and emission spectra of neat TQ1 and TQ1:PCBM blend films (Figure 1). For all experiments reported here the optical

EXPERIMENTAL SECTION

Sample Preparation. Films of TQ1 and TQ1 blended with PCBM in a 4:1 polymer:fullerene weight ratio were spin-cast (1000 rpm) from chloroform solution (10 mg mL−1) onto glass substrates. These films will be referred to as isotropic. Oriented films were obtained by manual rubbing of isotropic films at temperatures of 110 °C with a piece of velvet cloth, following the method reported in ref 21. To prevent the films from detaching from the substrate during the alignment procedure a thin interlayer of PEDOT:PSS (40 nm) was used between the substrate and the film. For consistency this interlayer was also present in isotropic films, and both isotropic and oriented films were heated at 110 °C for 1 min. The 4:1 blend ratio (excess of polymer) was required to obtain a sufficiently large degree of alignment of the polymer phase. 2D-POLIM Microscopy. Two-dimensional POLIM experiments were performed using a home-built wide-field fluorescence microscope based in a commercial Olympus IX7 inverted microscope using the 633 nm output of a He−Ne tunable laser. The laser light was passed through a cleanup filter and a wire-grid polarizer (Edmund Optics) before reaching the excitation polarization controller. The excitation controller is used to change the orientation of the linearly polarized light on the sample plane and consisted of an achromatic λ/2 plate for the vis (400−800 nm Thorlabs) mounted in a motorized rotation mount (Standa). Some of the optical elements placed between the excitation controller and the sample plane are birefringent and thus change the polarization state of the initially linearly polarized excitation light at the sample plane. Therefore, a berek compensator (New Focus) was placed after the excitation controller to introduce the opposite phase shift and recover the linear polarization at the sample plane (more details in the Supporting Information). A spatial filter was used to improve the quality of the excitation spot and to eliminate reflections introduced by the optics at the sample plane. A defocusing lens was used to obtain an excitation spot with diameter of 80 μm at the sample plane. A dry objective lens (40× Olympus, numerical

Figure 1. (A) Chemical structures of TQ1 and PCBM, with schematic indications for the donor excited state (blue) and the CT state (red). (B) Absorption (dashed lines) and fluorescence (full lines) of TQ1 (blue) and TQ1:PCBM blend films (red). The fluorescence spectrum of the blend films consists of residual TQ1 emission, centered at 740 nm, and a CT band, at 860 nm. The measured fluorescence intensity is largely affected by the gradual sensitivity drop of the photodetector (silicon CCD) for wavelength >800 nm, and thus, the CT emission band appears as a shoulder. The fluorescence spectra corrected for the spectral sensitivity of the setup is shown for comparison (dotted red line). All fluorescence spectra are scaled for clarity. The TQ1 and CT emission bands are broad and thus contribute in both spectral ranges of the emission filters. In blend films, the filter set polymer predominantly transmits emission from the TQ1 chromophores (∼90% of the total signal), while the filter set red allows a weak red tail of the TQ1 (30%) and all the interfacial CT state emission (70%) to reach the detector (more information in the Supporting Information).

densities of the neat and blend films were OD ≈ 0.25 at 620 nm. In absorption, both films peaked near the same wavelength (≈616 nm) and showed a similar overall spectral shape, with the TQ1:PCBM spectrum being slightly broader on the red edge. The fluorescence intensity of the blend films was approximately 100 times weaker than that of the neat films, which is attributed to quenching of the TQ1 fluorescence by the PCBM electron acceptor, yielding CT states and B

dx.doi.org/10.1021/cm502503f | Chem. Mater. XXXX, XXX, XXX−XXX

Chemistry of Materials

Article

subsequently free charge carriers.17,19,21 Moreover, the fluorescence spectrum of blend films contains, in contrast to neat films, an additional red-shifted band, which has been assigned to emission of the interfacial CT state.21 The spin-casting procedure resulted in a random organization of the TQ1 polymer phase within the film plane, and thus, we denote these films as isotropic. To investigate the relation of the CT state dipole moment to the orientation of the donor states, orientation of the polymer phase was accomplished by mechanical rubbing of the films against a velvet cloth at 110 °C (see Experimental Section). Hereafter these films are referred to as oriented. We used two different sets of emission filters (Figure 1) to distinguish between the photophysical properties of the polymer phase (filter set polymer) and the interfacial CT state (filter set red). All combined, we report the 2D polarization imaging (2D-POLIM) measurements at 8 different experimental conditions: neat TQ1 and TQ1:PCBM blend films × isotropic and aligned polymer phases × 2 spectral regions. Two-Dimensional Polarization Imaging. Neat and blend films were studied using 2D-POLIM to determine the relation between the absorbing and emissive states, and the excitation energy transfer processes occurring therein. 2D-POLIM is a powerful experimental method that can be seen as a further development of the fluorescence anisotropy and linear dichroism experiments. 2D-POLIM studies the correlation between the polarization of the sample emission and the electric field’s direction of the linearly polarized excitation light. In essence, the fluorescence intensity I of the sample is imaged for a set of combinations of the excitation and emission polarization angles, φex and φem. From this data we obtain a two-dimensional function so-called polarization portrait, I(φex, φem), for each pixel (of size 0.26 × 0.26 μm2) in the imaging plane (Figure 2). In its most general interpretation, the polarization portrait contains the emission polarization degree

at every excitation polarization angle.29,30 This two-dimensional function depends on the sample’s internal organization of chromophores and also on the EET between them. The polarization portrait is used to calculate the 2D-POLIM contrasts of each pixel, such as (i, ii) the modulation depths in excitation and emission, Mex and Mem, (iii, iv) the polarization phases, θex and θem, (v) the fluorescence anisotropy, r, and (vi) the energy funneling efficiency, ε. Mex and Mem describe the degree of orientation of the transition dipole moments, and conveniently map this onto the interval 0−1 (isotropic− uniaxial). θex and θem correspond to the main orientation axes of fluorescence excitation and emission, respectively. Their difference, the luminescence phase shift Δθ = θex − θem, indicates the presence of EET when Δθ ≠ 0. While the modulation depths and phases are obtained through integration of the polarization portrait (Supporting Information eqs S9 and S10), fluorescence anisotropy is calculated from two points in the polarization portrait (Supporting Information eq S11), namely, I∥ = (θex, θex) and I⊥= (θex, θex + π/2). r can be used to infer the presence or absence of EET; however, it is not wellsuited for quantitative measurements of the EET efficiency.31 On the other hand, analysis of the full polarization portrait via the so-called single f unnel approximation (SFA) allows us to quantify the EET processes of the system.30 In the SFA the energy funneling efficiency (ε) quantitatively describes the degree of EET from each dipole in the system toward a single emitter, which can have dipolar, elliptic, or isotropic character. The parameter ε ranges from 0, no transfer, to 1, full EET toward the single emitter. While the modulation depths and phases serve to characterize the sample’s internal chromophore geometry, 2D-POLIM contrasts can be used as indicators for the presence of EET in the form of Mem ≠ Mex, r smaller than the fundamental fluorescence anisotropy (r0), Δθ ≠ 0, and, crucially, through the direct quantification of EET as ε. To prepare ourselves for their interpretation in the measured polymer:fullerene blends we now shall briefly dwell on the conceptual behavior of Mex, Mem, r, and ε, in the context of isotropic and aligned ensembles of dipoles. Figure 3 depicts the qualitative behavior of the 2D-POLIM observables as functions of the degree of dipole alignment and EET obtained via a computer simulation (see Simulations of the CT Emission section for more details). Gradually increasing the degree of alignment in the dipole ensemble and in the absence of EET, the 2D-POLIM parameters develop as follows: (i) Mex increases monotonically toward 1, (ii) Mem = Mex, (iii) r is equal to the fundamental fluorescence anisotropy r0 and grows toward 1, and (iv) the energy funneling efficiency ε remains at zero. Allowing EET (i) does not affect Mex and (ii) slightly increases Mem relative to Mex for partially oriented samples; (iii) the observed r drops below r0 with a dynamic range that decreases with the degree of alignment, and (iv) ε increases toward one, being nearly independent of the degree of alignment. Note that the random orientation of the transition dipoles in isotropic samples leads to Mex = Mem = 0 and, in the absence of EET, to r0 = 0.4. The 2D-POLIM contrasts of each pixel are then used to generate new images of the sample plane. Figure 4 shows typical examples of modulation depth images obtained for isotropic and oriented TQ1:PCBM films (extra 2D-POLIM contrast images can be found in Supporting Information Figures S2−S9). Isotropic films are extremely uniform in their properties (see Figure 4A−C). However, this uniformity is lost

Figure 2. Exemplary polarization portrait and calculation of the 2DPOLIM parameters. The modulation depth and phase for fluorescence excitation (Mex and θex) and emission (Mem and θem) are obtained through the integration of the polarization portrait over φem or φex, respectively. Fluorescence anisotropy, r, is calculated using Supporting Information eq S11 with data points I∥ and I⊥. The energy funneling efficiency, ε, is calculated by analysis of the full polarization portrait via the single funnel approximation.30 C

dx.doi.org/10.1021/cm502503f | Chem. Mater. XXXX, XXX, XXX−XXX

Chemistry of Materials

Article

(Figure 4G). This is a perfect example of how 2D-POLIM reveals structural information that cannot be observed using normal fluorescence microscopy. Furthermore, this imaging data allows us to study correlations between different polarization parameters, since they were all obtained at the same time and for the same sample area.30 Of particular interest is the modulation correlation plot Mex versus Mem (Figure 5), since it compares

Figure 5. Modulation correlation plot for the TQ1:PCBM blend and neat TQ1 films. The data is presented as the normalized density for each specific experimental condition, in which all pixels of all images were considered. The information inside parentheses refers to the filters used. In the case of neat TQ1 films, isotropic and oriented, the density functions are indistinguishable between the red and polymer filters.

Figure 3. Schematic plots of 2D-POLIM contrasts as functions of polymer alignment and excitation energy transfer (see bottom pictograms and in-plot annotations). Shown are, from top to bottom, the modulation depths in excitation and emission, Mex and Mem, the anisotropy r, and the energy funneling efficiency ε. See Table 1 and text for details. Note that the degree of alignment is given in units of Mex.

the polarization properties of the absorbing and emitting states, which can differ from one another due to EET between differently polarized states. While for isotropic films the modulation depths in both excitation and emission are zero everywhere (see lower-left corner in Figure 5), the correlations

in oriented films, since the mechanical rubbing procedure induces micrometer-scale inhomogeneities (Figure 4D−F).32,33 Interestingly, for oriented films, the fluorescence intensity and polarization modulation images do not correlate completely

Figure 4. Exemplary 2D-POLIM images of isotropic TQ1:PCBM (top row) and oriented TQ1:PCBM (bottom row) films using the polymer filter. Contrasts are (A, D) fluorescence intensity, (B, E) Mex, and (C, F) Mem. (G) presents a 3 times zoom of the same area of D, E, and F (marked by the rectangles). From the zoom images we can appreciate the presence of structures in the polarization data (Mex or Mem) that are not visible in the intensity image. The scale bars are 26 μm long. Note that the fluorescence intensity images (A, D) reflect the shape of the exciting laser beam. D

dx.doi.org/10.1021/cm502503f | Chem. Mater. XXXX, XXX, XXX−XXX

Chemistry of Materials

Article

Table 1. Average Mex, Mem, r, and ε Values Obtained for TQ1 and TQ1:PCBM Filmsa composition

alignment

filter

Mex ± 0.01

Mem ± 0.03

EET ε

r ± 0.01

r0

TQ1

isotropic

polymer red polymer red polymer red polymer red

0.02 0.03 0.36 0.37 0.03 0.03 0.43 0.39

0.02 0.03 0.57 0.59 0.03 0.02 0.53 0.34

0.93 0.94 0.83 0.85 0.28 0.56 0.50 0.73

0.02 0.02 0.51 0.54 0.28 0.15 0.56 0.32

0.40 0.40 0.57 0.58 0.40 0.40 0.61 0.59

oriented TQ1:PCBM

isotropic oriented

Filter polymer transmits in the range 675−810 nm, while filter red transmit wavelengths >823 nm. r0 is the fundamental fluorescence anisotropy calculated for a set of nonrotating, noninteracting chromophores with Mex equal to that measured experimentally (Supporting Information Figure S10). a

quenching of TQ1 excitons by PCBM in the blend films. Let us consider the isotropic and oriented cases separately. (i) Those chromophores that are oriented along the electric field vector of the incident light are preferentially excited (photoselection). In isotropic neat films, this initial anisotropic population of excited chromophores is completely lost in fluorescence emission (r ≈ 0 and ε ≈ 1, Supporting Information Figure S11). This is the result of extremely efficient EET (see effect of increased EET for Mex = 0 in Figure 3) between differently oriented and energetically similar TQ1 chromophores (homo-EET). On the contrary, the TQ1 fluorescence emission in the blends is subject to strong photoselection (large r ≈ 0.3 and small ε ≈ 0.3, Supporting Information Figure S11) indicating lower homo-EET efficiency between differently oriented chromophores. Moreover, timeresolved fluorescence anisotropy measurements in neat films of conjugated polymers have shown ultrafast anisotropy decays (from r0 = 0.4 to r ≈ 0.35 for MEH-PPV) during the first ≈200 fs, followed by a significantly slower decay.28,34 These fast and slow decays are attributed to exciton relaxation and EET, respectively.35,36 Therefore, the decay from r0 to r = 0.3 (and ε ≈ 0.3) observed in blend films is probably not the result of homo-EET but rather due to the ultrafast exciton relaxation of the TQ1 chromophores, which is accompanied by a change of the transition dipole moment orientation. (ii) In oriented films of both the neat and blend material, the TQ1 phase has Mem values larger than Mex (Figure 5). However, from the ε values, the differences between Mem and Mex, and the departure of r from r0, we infer a more efficient EET among TQ1 chromophores in neat TQ1 than in blend films (see Figure 3). Mem > Mex is the result of selective EET toward a subset of chromophores that are more aligned than those responsible for the fluorescence excitation. This behavior has been observed before in single multichromophoric systems, such as aggregates and conjugated polymers.29,37−45 Considering the previously described observations we can conclude that the rate of CT formation competes with the homo-EET rate within the TQ1 phase, to the extent that it renders EET processes among TQ1 chromophores inefficient. In other words, the interaction with the PCBM acceptor reduces the lifetime of the excited states of TQ1 chromophores and limits the time available for EET to occur between differently oriented TQ1 chromophores. This is in agreement with previous time-resolved measurements of TQ1:PCBM films,46 which showed the formation of the CT state in ≈300 fs and EET times of several picoseconds in the polymer, and with data obtained for other polymer/fullerene blends.47,48

become more insightful in the case of the oriented films, as we will discuss further below. As a consequence of the broad emission spectrum of TQ1 at room temperature, the emission from the TQ1 phase in both spectral regions transmitted by the polymer and red filter arises from the same set of TQ1 states. Therefore, the 2D-POLIM contrasts for neat TQ1 films are independent of the emission filter used (see Table 1), and the Mex versus Mem correlations observed under both filters are indistinguishable (see Figure 5). In order to obtain statistically significant results, for each experimental condition we measured at least a total of 7 different areas coming from 2 or more samples. In Table 1 we report the average values of the relevant 2D-POLIM contrasts over all pixels of all the images for each experimental condition. The average values provide an overview and will serve as the starting point of the discussion. 2D-POLIM of the TQ1 Phase. We will first direct our attention toward the study of the TQ1 phase in neat and blend films, as we need this knowledge to understand the relationship between TQ1 and PCBM. Therefore, in the following paragraphs we discuss the results of neat TQ1 and TQ1:PCBM films using the polymer filter (see Table 1). In isotropic films, the TQ1 transition dipole moments for fluorescence excitation and emission are randomly oriented on the polymer phase for both neat TQ1 and TQ1:PCBM blends (Mex ≈ Mem ≈ 0). This behavior is extremely uniform throughout the isotropic films, as we can observe in the 2DPOLIM images and Mex versus Mem correlation plot (Figures 4B,C and 5). The alignment procedure orients the TQ1 chromophores (Mex > 0, Figure 4E) toward the alignment direction (as indicated by θex, Supporting Information Figures S6−S9).21,32,33 Although the degree of alignment (Mex) is not uniform through the film, the mechanical rubbing creates elongated features of uniform Mex (approximate size 5 × 60 μm2, Figure 4E). As expected, the emission dipole moments are also oriented by the alignment procedure (Mem > 0, Figure 4F) and have the same average orientation direction as the absorbing dipoles, ⟨Δθ⟩ = 0 (see Supporting Information Figures S6−S9 for Δθ histograms). However, to understand the relation between Mex and Mem we must consider the EET processes occurring within the polymer phase. For both isotropic and oriented films, the EET among TQ1 chromophores is considerably larger in neat TQ1 films than in TQ1:PCBM blends. This is demonstrated by the lower values of r (and larger values of ε) for the neat films. This occurs despite neat and blend films having a similar orientation degree of the polymer phase (Mex values using the polymer filter). The reason for this completely different EET behavior is the E

dx.doi.org/10.1021/cm502503f | Chem. Mater. XXXX, XXX, XXX−XXX

Chemistry of Materials

Article

We suggest that the ε measured for the blend films indicates that the donor and acceptors are very well-mixed. In blends, the excitons created in the polymer at distances (RDA) further than ca. 1 nm from the PCBM acceptor have to migrate toward the interface in order to be quenched. If excitons in TQ1 were generated at RDA > 10 nm they would have only a small chance to interact with the PCBM acceptor due to the limited exciton migration in CPs, 49 which occurs through a hopping mechanism.50 Residual polymer fluorescence from such diffusing excitons must have a strong homo-EET signature, similar to that observed in neat films (ε ≈ 1). Note that because 2D-POLIM is based on fluorescence, it is extremely sensitive to such nonquenched TQ1 chromophores, if they were present. Further, as mentioned above, exciton relaxation in the polymer phase gives rise to ε > 0, without the need of homo-EET processes. Therefore, we suggest that the intermediate value of ε for the polymer phase in the blends indicates that most of the donor chromophores are located at RDA < 10 nm. 2D-POLIM of the CT Emission. In the following section we will concentrate on obtaining information about the emissive CT states, in particular its relation to the orientation of the donor states, and its population and decay processes. In blends with high polymer to fullerene ratios, as the one used in this study, free charge generation is preceded by the formation of bound CT states.7 This is because fullerene clusters and thus delocalized states in the acceptor domain that are energetically resonant and coupled to the donor excitons should be rare. Therefore, in the following discussion we will ignore the presence of direct, high energy channels for charge dissociation that bypass the CT states. In a simple perturbation theoretical treatment24,51,52 the transition dipole moment of the emissive CT state (MCT) can be approximated as a superposition of the pure ionic state and the electronic configurations of the locally excited donor and acceptor. In this work we consider MCT to consist of two components: first, a transition dipole moment MCT∥ which is “borrowed” from the donor (contribution of the locally excited donor) and is thus oriented parallel to its transition dipole moment; and second, the dipole moment MCTr, which represents the contributions of (i) the ionic (charge-separated) state with the ground state of the complex, and (ii) the locally excited acceptor. The full dipole moment of the emissive CT state is the vectorial sum of these two,10,21 to wit: r MCT = MCT + MCT

Figure 6. Top: The CT states are populated through the TQ1 absorption (kCT). Only those TQ1 chromophores that are close to the interface are directly coupled to the CT states. TQ1 chromophores in the bulk must transfer their excitation energy toward the interface to populate the CT states. Most of the populated CT states are nonemissive due to efficient charge separation and nonradiative decay to the ground state (knrCT, nonradiative decay). However, some CT states are able to live long enough to emit fluorescence. These are the CT states that we measure in our experiments. Bottom: The fluorescence excitation modulation depth of the TQ1 (polymer filter) is larger than that of the CT states (red filter).

proportional to the yield of CT formation kCT, and to the fluorescence quantum yield ΦCT of the CT state (Figure 6). In oriented blend films the Mex values observed using the polymer filter (Mex = 0.43) exceed those observed for the red filter (Mex = 0.39), by an amount considerably larger than our error. This is observed not only in the overall averages, but also holds for each pixel separately (Figure 6, bottom). Mex is defined by the degree of orientation of the transition dipole moments that contribute to the fluorescence excitation cross section, σex = σ × Φ, where σ is the absorption cross section and Φ the fluorescence quantum yield. Mex is usually interpreted only in terms of the orientation of the transition dipole moments. However, the oriented blend films were excited through the same TQ1 absorption dipoles, regardless of the emission filter used. Therefore, the Mex difference between the polymer and red filters cannot be related to the orientation of the TQ1 dipoles or their σ. The only difference between these two experiments is the state responsible for the fluorescence emission: TQ1 (polymer filter) or CT emission (red filter). Considering the indirect excitation of the CT states (kCT) and assuming that the Φ of TQ1 does not depend on their orientation, we can conclude that the CT states populated through the excitation of more oriented TQ1 domains have lower luminescence intensity. This can be interpreted as a reduced kCT and/or reduced ΦCT of the more oriented domains, in comparison with CT states populated through the more isotropic regions of the film.

(1) r

The orientation of MCT depends on the directions of the ionic state and the fullerene transition dipole moments. The orientation of the fullerene acceptors is expected to be unaffected by the alignment procedure due to their shape. Therefore, we assume that MCTr is randomly oriented.21 This principle is used in our polarization sensitive experiments to yield insight into the ratio of the average vector magnitudes MCT∥/MCTr, which will be compared to previous studies. In all our experiments, we always excite the TQ1 absorption dipoles. Therefore, CT states are indirectly populated through the excitation of TQ1 chromophores (Figure 6). The later can be (i) located at the interface and directly coupled to CT states, or (ii) located away from the interface and indirectly coupled to CT states, through EET toward the interface. On the other hand, we expect most CT states to be nonfluorescent due to efficient charge separation, which acts as a nonradiative decay. Thus, only relatively few CT states live long enough to be emissive. The luminescence intensity of a CT state is F

dx.doi.org/10.1021/cm502503f | Chem. Mater. XXXX, XXX, XXX−XXX

Chemistry of Materials

Article

Reduced population of CT states (smaller kCT) could be the result of local morphology. In the case that the blend does not phase separate, a smaller concentration of PCBM acceptors in more oriented TQ1 domains would reduce the amount of CT states (Figure 7A). For example, orientation of the polymer

results we cannot disentangle their contributions. However, it is known that devices with annealing induced crystallization of the polymer phase are superior in terms of current extraction, compared to the initially cast films.17 This is probably the result of enhanced hole mobility in crystalline domains of the polymer phase. In terms of nonradiative pathways, enhanced hole mobility reduces the chance of geminate charge recombination, which is beneficial for charge separation. Therefore, we suggest that the larger hole mobility of the more oriented nanometerscale domains of the polymer phase contribute to the more efficient quenching of the CT emission. We now turn our attention to the modulation in emission, Mem, which depends on the orientation and population of the emissive CT states. As described above, the CT state dipole moment can be represented by the vector sum of two components: MCT∥ and MCTr. For isotropic blends, both CT components are randomly oriented, and thus, Mem = 0. Additionally, EET processes do not have the potential to induce an anisotropic CT state population. For oriented films, the TQ1 chromophores are preferentially oriented, and so are the MCT∥ components. As a result, the CT dipole moments are preferentially oriented to a degree that depends on the ratio MCTr/MCT∥, which, unless MCTr ≫ MCT∥, leads to Mem > 0. To fully comprehend the relationship between the Mem and the CT components we must consider the possibility of EET processes occurring within the system, which we will do in the following. For CT state emission, the Mex versus Mem correlation, anisotropy r, and the EET efficiency ε all depend on the relative orientation of the initially excited TQ1 chromophores (photoselected by the linearly polarized excitation) and the subsequently emissive CT states. This relative angle not only depends on the randomly oriented CT component MCTr, but also is potentially affected by (i) homo-EET in the TQ1 phase prior to CT formation, and (ii) homo-EET between CT states. The presence of randomly oriented MCTr reduces the anisotropy of the CT emission relative to the initially photoselected TQ1 dipoles, and, thus, is expected to yield ε > 0. In isotropic films, any efficient EET among CT states would very rapidly reduce the anisotropy toward r ≈ 0 and increase ε ≈ 1, which lies in contrast to our observations (r = 0.15 and ε = 0.56). This suggests that the homo-EET between differently oriented CT states is not efficient. On the other hand, the CT emission of oriented blends present a very unusual Mex versus Mem correlation, with 0 < Mem < Mex (see Figure 5), signaling the presence of EET toward less oriented emissive sites. To the best of our knowledge, this is the first multichromophoric system in which this behavior was observed. EET processes in multichromophoric systems generally tend to increase Mem relative to Mex,29,37−45 as we observed previously for the TQ1 phase. Combining this with the results from isotropic blend films, the unusual Mex versus Mem correlation indicates that the change in orientation between absorbing TQ1 dipoles and emitting CT states is mainly due to the MCTr component. Simulations of the CT Emission. To comprehend the full picture of the interfacial CT state, accounting for all potential EET processes, degree of orientation of the TQ1 phase, mixture between TQ1 and CT emission, as well as the ratio MCTr/MCT∥, we performed 2D-POLIM analyses on computer simulated data. We used a computer model to simulate the fluorescence intensity of neat and blend films. Here, the polymer phase emission arises from a large ensemble of dipole moments (described in full detail in the Supporting

Figure 7. Possible origins of lower CT luminescence intensity from oriented regions of the polymer phase. (A) The blend does not phase separates but the concentration of the PCBM acceptors is smaller in the oriented domains of the polymer. (B) The blend phase separates, and the degree of orientation of the polymer phase depends on the distance to the interface. (C, D) All degrees of orientation are equally represented at the interface. However, the nonradiative rates of the CT states depend on the degree of orientation of the polymer at the interface.

phase may induce the formation of nanometer-scale highly oriented “crystalline” domains that could exclude PCBM, as observed in P3HT.53 On the other hand, if the blend phase separates, lower kCT could be related to a larger degree of alignment of the polymer phase in the bulk than close to the interface (Figure 7B). As a result, the energy absorbed in the more oriented TQ1 domains would have to overcome a larger distance to populate the interfacial CT states, thus decreasing their kCT. An alternative scenario is that all degrees of orientation of the donor are equally represented at the interface, and thus equally populate the CT states (Figure 7C,D). In this case, we have to assume that the reduced CT luminescence arises from ΦCT which is dependent on the degree of orientation of the polymer phase at the interface. This conclusion is in agreement with previous observations in polymer:fullerene blends where annealing of the films induced a crystallization of the polymer phase, accompanied by significantly lower CT state EL than before annealing. This observation was attributed to an actual quenching of the radiative CT state emission.17 In this context, there are two main nonradiative pathways that must be considered: (i) the nonradiative coupling of the CT state to the ground state and (ii) dissociation of the CT state (charge separation, Figure 6). The interplay between these two nonradiative processes is extremely important to solar cell operation. Unfortunately, on the basis of our experimental G

dx.doi.org/10.1021/cm502503f | Chem. Mater. XXXX, XXX, XXX−XXX

Chemistry of Materials

Article

the emissive CT state borrows dipole moment from the donor with a stronger coupling than thought previously. The values for electronic coupling between excited donor and CT state found in ref 21 are thus underestimated by a factor of ≈2, and we revise them for the TQ1:fullerene system up to 12 meV. The main difference between our current approach and that of ref 21 is that we account for (i) the homo-EET and exciton relaxation in the polymer phase prior to CT formation, and (ii) the contribution of the residual (but significant) polymer emission tail, which spectrally coincides with the CT emission band. We shall briefly dwell on the apparent absence of EET between differently oriented emissive CT states. This can happen if the distance between such states is substantially larger than the Förster radius. Förster radius is proportional to the 6th power root of the oscillator strength (dipole moment squared) of the acceptor multiplied by the fluorescence quantum yield of the donor, where both the donor and acceptor are emissive CT states. By comparing the CT and polymer (TQ1) absorption of blends with low polymer concentration, the average oscillator strength of a CT state has been estimated to be about 1/300 of the oscillator strength of the TQ1 spectroscopic unit,21 which is consistent with calculations on other polymer:fullerene systems.26,27 Note that this estimation gives the oscillator strength averaged over all CT states (emissive and nonemissive), while emissive CT states probably have much larger oscillator strengths than those which undergo ultrafast charge separation on subpicosecond time scale. The lifetime of CT emission is known to be in the nanosecond range,14,54 which means that if emissive CT states separate to charges at all, it takes orders of magnitude longer time than for nonemissive CT states. Using EL measurements, previous works have found that CT state absorption is able to efficiently generate free charges with the same yield as an exciton in the donor phase.55 However, the quantum yield of CT states’ EL is 2 orders of magnitude lower than that of PL (10−6 and 10−4, respectively). PL, which is studied in our work, originates preferentially from the parts of the film that have lower charge separation efficiency. Therefore, PL probes mainly those CT states which do not make it to free carriers but instead mostly recombine geminately. In a device, geminate recombination occurs at places which are poorly connected to electrodes. In contrast, EL probes CT states formed via recombination of free carriers at interfaces which are connected to electrodes. Due to the 2 orders of magnitude difference in the luminescence quantum yield, our experiments (PL) are much less sensitive to the recombination of free charges than that of geminate pair recombination. Unfortunately, since we do not know the luminescence quantum yield of the emissive CT states, we still cannot estimate the Förster radius. Nevertheless, due to the 6th power root dependency, even if the product of fluorescence quantum yield and oscillator strength is 10 000 times smaller than that for TQ1, the Förster radius will decrease only by a factor of 4.6 in comparison to that in a pure polymer phase (about 5−7 nm49). So, we can still expect energy transfer over ca. 1.5 nm between the emissive CT states. Since we do not observe it, the distances between the emissive CT states must be more than 1.5 nm. This is in agreement with the idea that only relatively few CT states are emissive, which results in large distances between the physical locations of these states. It should be noted that these

Information, section 5). We generated ensembles of dipoles with isotropic and oriented angular distributions reflecting the Mex of the experimental films. Within this ensemble we allowed homo-EET in the form of a hopping process, with a rate (kTQ1EET) which depends on the relative orientation of dipoles. kTQ1EET was chosen such that the simulated r, Mem, and ε reproduced the experimental observations for pure TQ1 emission. To consider the formation of the CT state, we limited the time available for the EET processes among the ensemble dipoles, using a parameter which can be tuned arbitrarily. We found that this approach yields 2D contrasts consistent with the experimental results for the polymer phase emission in blends. In particular, it turns out that the rate of CT formation far exceeds the rate of homo-EET, kCT ≫ kTQ1EET. To enable CT emission, each absorption dipole of the ensemble was granted a second emission dipole constructed as the weighted vectorial sum of the absorbing dipole (MCT∥) and a randomly oriented dipole (MCTr). To simulate our experimental conditions, 30% of the emission was taken from states belonging to the initial set of absorption dipoles, while 70% was taken from the simulated CT states (Figure 1). Within the absorption dipoles we allowed homo-EET rates of kTQ1‑isoEET and kTQ1‑oriEET as obtained previously for the isotropic and oriented polymer phase emission, respectively. Our model confirmed that EET among emissive CT states is highly inefficient: upon introduction of such transfer processes, both r and Mem quickly drop to zero and ε reaches 1, even for comparatively low EET rates, contrary to our experimental results. Figure 8 shows the evolution of the 2D-POLIM contrasts as a function of the ratio MCTr/MCT∥, for simulations of both isotropic and oriented blends, where EET is allowed among donors, but not among CT states. We found that a ratio of MCTr/MCT∥ ≈ 0.85 provides a close match to the experimental observations. This ratio is significantly smaller than purely anisotropy-derived results (1.65 ± 0.45),21 which means that

Figure 8. Simulated 2D-POLIM contrasts Mex (blue), Mem (green), r (black), and ε (red) as a function of the ratio MCTr/MCT∥ for CT emission of an isotropic ensemble (solid lines) and an oriented ensemble of dipoles (dashed lines). For the latter, the degree of orientation was tuned to fulfill the experimental observation of Mex = 0.39. There is no homo-EET between CT dipoles, while homo-EET among TQ1 dipoles is set to kTQ1‑oriEET for the oriented films and kTQ1‑isoEET for isotropic films. The gray interval indicates for which MCTr/MCT∥ the simulation roughly corresponds to experimental observations (highlighted by ellipses of corresponding color). H

dx.doi.org/10.1021/cm502503f | Chem. Mater. XXXX, XXX, XXX−XXX

Chemistry of Materials



distances will strongly depend on the morphology of the blend and the properties of the donor/acceptor interface. In the preceding discussion we have so far ignored the potential for EET between emissive and nonemissive CT states. This should be considered, since these states have similar energies, small spatial separation, and possibly nonvanishing dipole moments.51 We cannot detect this type of transfer, since our signal arises from luminescence, and this transfer is a quenching mechanism. On the other hand, we should be able to detect the back-transfer from nonemissive to emissive CT states. Therefore, our observation of inefficient EET among CT states also implies that there is no efficient back-transfer from nonemissive to emissive CT states. This is consistent with the idea that the predominant mechanism depopulating nonemissive CT states is fast charge separation.

AUTHOR INFORMATION

Corresponding Author

*E-mail: [email protected]. Notes

The authors declare no competing financial interest.



ACKNOWLEDGMENTS This study was financially supported by The Swedish Research Council, the Knut & Alice Wallenberg Foundation, Crafoordska Stiftelsen, and Linnaeus Grant to Lund Laser Center. O.I. thanks the Wallenberg foundation for a Scholar grant.



REFERENCES

(1) Sariciftci, N. S.; Smilowitz, L.; Heeger, A. J.; Wudl, F. Science 1992, 258, 1474. (2) Yu, G.; Gao, J.; Hummelen, J. C.; Wudl, F.; Heeger, A. J. Science 1995, 270, 1789. (3) Campoy-Quiles, M.; Ferenczi, T.; Agostinelli, T.; Etchegoin, P. G.; Kim, Y.; Anthopoulos, T. D.; Stavrinou, P. N.; Bradley, D. D. C.; Nelson, J. Nat. Mater. 2008, 7, 158. (4) Chen, H.; Hou, J.; Zhang, S.; Liang, Y.; Yang, G.; Yang, Y.; Yu, L.; Wu, Y.; Li, G. Nat. Photonics 2009, 3, 649. (5) Park, S. H.; Roy, A.; Beaupre, S.; Cho, S.; Coates, N.; Moon, J. S.; Moses, D.; Leclerc, M.; Lee, K.; Heeger, A. J. Nat. Photonics 2009, 3, 297. (6) Liang, Y.; Xu, Z.; Xia, J.; Tsai, S.-T.; Wu, Y.; Li, G.; Ray, C.; Yu, L. Adv. Mater. 2010, 22, E135. (7) Savoie, B. M.; Rao, A.; Bakulin, A. A; Gelinas, S.; Movaghar, B.; Friend, R. H.; Marks, T. J.; Ratner, M. A. J. Am. Chem. Soc. 2014, 136, 2876. (8) Gélinas, S.; Rao, A.; Kumar, A.; Smith, S. L.; Chin, A. W.; Clark, J.; van der Poll, T. S.; Bazan, G. C.; Friend, R. H. Science 2014, 343, 512. (9) Gao, F.; Inganäs, O. Phys. Chem. Chem. Phys. 2014, 16, 20291. (10) Vandewal, K.; Tvingstedt, K.; Inganäs, O. Semicond. Semimetals 2011, 85, 261−293. (11) Benson-Smith, J. J.; Goris, L.; Vandewal, K.; Haenen, K.; Manca, J. V.; Vanderzande, D.; Bradley, D. D. C.; Nelson, J. Adv. Funct. Mater. 2007, 17, 451. (12) Loi, M. A.; Toffanin, S.; Muccini, M.; Forster, M.; Scherf, U.; Scharber, M. Adv. Funct. Mater. 2007, 17, 2111. (13) Hallermann, M.; Haneder, S.; Da Como, E. Appl. Phys. Lett. 2008, 93, 053307. (14) Veldman, D.; Ipek, O.; Meskers, S. C. J.; Sweelssen, J.; Koetse, M. M.; Veenstra, S. C.; Kroon, J. M.; van Bavel, S. S.; Loos, J.; Janssen, R. A. J. J. Am. Chem. Soc. 2008, 130, 7721. (15) Gonzalez-Rabade, A.; Morteani, A. C.; Friend, R. H. Adv. Mater. 2009, 21, 3924. (16) Zhou, Y.; Tvingstedt, K.; Zhang, F.; Du, C.; Ni, W.-X.; Andersson, M. R.; Inganäs, O. Adv. Funct. Mater. 2009, 19, 3293. (17) Tvingstedt, K.; Vandewal, K.; Gadisa, A.; Zhang, F.; Manca, J.; Inganäs, O. J. Am. Chem. Soc. 2009, 131, 11819. (18) Veldman, D.; Meskers, S. C. J.; Janssen, R. a. J. Adv. Funct. Mater. 2009, 19, 1939. (19) Tvingstedt, K.; Vandewal, K.; Zhang, F.; Inganäs, O. J. Phys. Chem. C 2010, 114, 21824. (20) Jarzab, D.; Cordella, F.; Gao, J.; Scharber, M.; Egelhaaf, H.-J.; Loi, M. A. Adv. Energy Mater. 2011, 1, 604. (21) Vandewal, K.; Tvingstedt, K.; Inganäs, O. Phys. Rev. B 2012, 86, 035212. (22) Baran, D.; Li, N.; Breton, A.-C.; Osvet, A.; Ameri, T.; Leclerc, M.; Brabec, C. J. J. Am. Chem. Soc. 2014, 136, 10949. (23) Vandewal, K.; Tvingstedt, K.; Gadisa, A.; Inganäs, O.; Manca, J. V. Phys. Rev. B 2010, 81, 125204. (24) Huang, Y.; Westenhoff, S.; Avilov, I.; Sreearunothai, P.; Hodgkiss, J. M.; Deleener, C.; Friend, R. H.; Beljonne, D. Nat. Mater. 2008, 7, 483.



CONCLUSIONS 2D-POLIM is a powerful technique which is ideally suited for the study of anisotropic systems, such as the oriented polymer:fullerene films discussed here. The key advantage of our technique is the ability to quantify EET directly through the energy funneling efficiency, ε. We found that efficient EET among TQ1 is strongly reduced due to competing CT state formation. Further, we argue that the reduced EET in the polymer phase of blend films indicates that donor and acceptor phases are well-mixed, and estimate that excitons need to migrate much less than 10 nm in order to reach the interface. We found that the more oriented the polymer phase is, the weaker the CT luminescence becomes. This effect could be explained by different morphological scenarios, or, alternatively, by an increase in the charge-separation rate due to improved hole-mobility in oriented donor domains. We conducted a range of computer simulations to support our understanding of the photophysical properties of the emissive CT state. We were able to determine that our experimental observations were best fit by a CT state geometry with the ratio MCTr/MCT∥ ≈ 0.85. This differs from the previously found value for TQ1:PCBM roughly by a factor of 2, which indicates that emissive CT states borrow more dipole moment from the polymer donor than previously thought. We revise the values for electronic coupling between excited donor and CT state to 12 meV for the TQ1:PCBM system. Notably, we find that there is no efficient EET among emissive CT states. We argue that this arises as a consequence of the large spatial distances between the emissive CT states. When considering the possibility of EET between emissive and nonemissive CT states, it becomes clear that the back-transfer from nonemissive to emissive states is absent in our observations. This suggests that the observed emissive CT states are spatially or energetically isolated from other CT states.



Article

ASSOCIATED CONTENT

S Supporting Information *

Ratio of the emission intensities for the TQ1 and CT state in the spectral regions of the emission filters. Polarization portraits and calculation of the 2D-POLIM contrasts. Extra 2D-POLIM images for all experimental conditions. Relationship between the fundamental anisotropy and the fluorescence excitation modulation depth. Details about the simulation of the CT emission. This material is available free of charge via the Internet at http://pubs.acs.org/. I

dx.doi.org/10.1021/cm502503f | Chem. Mater. XXXX, XXX, XXX−XXX

Chemistry of Materials

Article

(25) Kanai, Y.; Grossman, J. C. Nano Lett. 2007, 7, 1967. (26) Yi, Y.; Coropceanu, V.; Brédas, J.-L. J. Mater. Chem. 2011, 21, 1479. (27) Liu, T.; Troisi, A. J. Phys. Chem. C 2011, 115, 2406. (28) Grage, M.; Zaushitsyn, Y.; Yartsev, A.; Chachisvilis, M.; Sundström, V.; Pullerits, T. Phys. Rev. B 2003, 67, 205207. (29) Mirzov, O.; Bloem, R.; Hania, P. R.; Thomsson, D.; Lin, H.; Scheblykin, I. G. Small 2009, 5, 1877. (30) Camacho, R.; Thomsson, D.; Yadav, D.; Scheblykin, I. G. Chem. Phys. 2012, 406, 30. (31) Sun, Y.; Wallrabe, H.; Seo, S.-A.; Periasamy, A. ChemPhysChem 2010, 462. (32) Lee, S. W.; Chae, B.; Lee, B.; Choi, W.; Kim, S. B.; Kim, S. I.; Park, S.-M.; Jung, J. C.; Lee, K. H.; Ree, M. Chem. Mater. 2003, 15, 3105. (33) Biniek, L.; Leclerc, N.; Heiser, T.; Bechara, R.; Brinkmann, M. Macromolecules 2013, 46, 4014. (34) Dykstra, T. E.; Hennebicq, E.; Beljonne, D.; Gierschner, J.; Claudio, G.; Bittner, E. R.; Knoester, J.; Scholes, G. D. J. Phys. Chem. B 2009, 113, 656. (35) Collini, E.; Scholes, G. D. Science 2009, 323, 369. (36) Hwang, I.; Scholes, G. D. Chem. Mater. 2011, 23, 610. (37) Müller, J. G.; Lupton, J. M.; Feldmann, J.; Lemmer, U.; Scherf, U. Appl. Phys. Lett. 2004, 84, 1183. (38) Becker, K.; Lupton, J. M. J. Am. Chem. Soc. 2006, 128, 6468. (39) Becker, K.; Lupton, J. M.; Feldmann, J.; Setayesh, S.; Grimsdale, A. C.; Müllen, K. J. Am. Chem. Soc. 2006, 128, 680. (40) Lin, H.; Tabaei, S. R.; Thomsson, D.; Mirzov, O.; Larsson, P.O.; Scheblykin, I. G. J. Am. Chem. Soc. 2008, 130, 7042. (41) Lin, H.; Hania, R. P.; Bloem, R.; Mirzov, O.; Thomsson, D.; Scheblykin, I. G. Phys. Chem. Chem. Phys. 2010, 12, 11770. (42) Traub, M. C.; Lakhwani, G.; Bolinger, J. C.; Vanden Bout, D.; Barbara, P. F. J. Phys. Chem. B 2011, 115, 9941. (43) Tian, Y.; Camacho, R.; Thomsson, D.; Reus, M.; Holzwarth, A. R.; Scheblykin, I. G. J. Am. Chem. Soc. 2011, 133, 17192. (44) Tubasum, S.; Camacho, R.; Meyer, M.; Yadav, D.; Cogdell, R. J.; Pullerits, T.; Scheblykin, I. G. Phys. Chem. Chem. Phys. 2013, 15, 19862. (45) Camacho, R.; Thomsson, D.; Sforazzini, G.; Anderson, H. L.; Scheblykin, I. G. J. Phys. Chem. Lett. 2013, 4, 1053. (46) Vithanage, D. A.; Wang, E.; Wang, Z.; Ma, F.; Inganäs, O.; Andersson, M. R.; Yartsev, A.; Sundström, V.; Pascher, T. Adv. Energy Mater. 2014, 4, 1301706. (47) Janssen, R. A. J.; Hummelen, J. C.; Sariciftci, N. S. MRS Bull. 2011, 30, 33. (48) Nelson, J. Curr. Opin. Solid State Mater. Sci. 2002, 6, 87. (49) Shaw, P. E.; Ruseckas, A.; Samuel, I. D. W. Adv. Mater. 2008, 20, 3516. (50) Scholes, G. D. Annu. Rev. Phys. Chem. 2003, 54, 57. (51) Bixon, M.; Jortner, J.; Verhoeven, J. W. J. Am. Chem. Soc. 1994, 7349. (52) Gould, I. R.; Young, R. H.; Mueller, L. J.; Albrecht, A. C.; Farid, S. J. Am. Chem. Soc. 1994, 116, 8188. (53) Swinnen, A.; Haeldermans, I.; vande Ven, M.; D’Haen, J.; Vanhoyland, G.; Aresu, S.; D’Olieslaeger, M.; Manca, J. Adv. Funct. Mater. 2006, 16, 760. (54) Pal, S. K.; Kesti, T.; Maiti, M.; Zhang, F.; Inganäs, O.; Hellström, S.; Andersson, M. R.; Oswald, F.; Langa, F.; Osterman, T.; Pascher, T.; Yartsev, A.; Sundström, V. J. Am. Chem. Soc. 2010, 132, 12440. (55) Vandewal, K.; Albrecht, S.; Hoke, E. T.; Graham, K. R.; Widmer, J.; Douglas, J. D.; Schubert, M.; Mateker, W. R.; Bloking, J. T.; Burkhard, G. F.; Sellinger, A.; Fréchet, J. M. J.; Amassian, A.; Riede, M. K.; McGehee, M. D.; Neher, D.; Salleo, A. Nat. Mater. 2014, 13, 63.

J

dx.doi.org/10.1021/cm502503f | Chem. Mater. XXXX, XXX, XXX−XXX