• No results found

University of Groningen Organic donor-acceptor systems Serbenta, Almis

N/A
N/A
Protected

Academic year: 2022

Share "University of Groningen Organic donor-acceptor systems Serbenta, Almis"

Copied!
35
0
0

Bezig met laden.... (Bekijk nu de volledige tekst)

Hele tekst

(1)

Organic donor-acceptor systems Serbenta, Almis

IMPORTANT NOTE: You are advised to consult the publisher's version (publisher's PDF) if you wish to cite from it. Please check the document version below.

Document Version

Publisher's PDF, also known as Version of record

Publication date:

2016

Link to publication in University of Groningen/UMCG research database

Citation for published version (APA):

Serbenta, A. (2016). Organic donor-acceptor systems: Charge generation and morphology. University of Groningen.

Copyright

Other than for strictly personal use, it is not permitted to download or to forward/distribute the text or part of it without the consent of the author(s) and/or copyright holder(s), unless the work is under an open content license (like Creative Commons).

The publication may also be distributed here under the terms of Article 25fa of the Dutch Copyright Act, indicated by the “Taverne” license.

More information can be found on the University of Groningen website: https://www.rug.nl/library/open-access/self-archiving-pure/taverne- amendment.

Take-down policy

If you believe that this document breaches copyright please contact us providing details, and we will remove access to the work immediately and investigate your claim.

Downloaded from the University of Groningen/UMCG research database (Pure): http://www.rug.nl/research/portal. For technical reasons the number of authors shown on this cover page is limited to 10 maximum.

Download date: 15-10-2022

(2)

67

Chapter 3

Bulk Heterojunction morphology of polymer:fullerene blends revealed by ultrafast spectroscopy

Part of this chapter is published in Ultrafast Phenomena XIX Proceedings of the 19th International Conference, Okinawa Convention Center, Okinawa, Japan, July 7-11, 2014:

Maxim S. Pshenichnikov, Almis Serbenta, and Paul H. M. van Loosdrecht.

The material of this chapter is submitted to journal:

Almis Serbenta, Oleg V. Kozlov, Paul H.M. van Loosdrecht and Maxim S.

Pshenichnikov

3. Abstract

Morphology of organic photovoltaic blends is one of the key factors influencing power conversion efficiency of organic solar cells. We demonstrate that visible pump – infrared (IR) probe spectroscopy is a powerful tool to investigate the morphology of the polymer:fullerene blends on the fly. The soluble fullerene derivative PC71BM was used as a light absorber while the appearance of charges on the polymer was probed with MID-IR pulses. The essence of the proposed method is the observation of the appearance of charges on the polymer, which is directly related to exciton diffusion, inside and out of PC71BM clusters. The details of the PC71BM exciton dissociation dynamics provide the necessary information to estimate the characteristic size of PC71BM clusters. Our studies demonstrate noticeable differences in the polymer:fullerene mixing behavior for different polymers. These results have direct implication for organic solar cell design where fullerene derivatives are the key players.

3.1. Introduction

Organic materials have recently gained interest for their potential use in renewable energy technology, energy efficient lighting applications, and in other optoelectronic devices1. Photophysical properties of organic materials strongly depend on their molecular packing and nanoscale structure, called morphology2. Morphology, in particular, plays a crucial role in organic photovoltaics where charge extraction is the final goal1, 3-5. Light absorption in organic materials typically creates strongly bound electron-hole pairs, called the Frenkel excitons6. These excitons are separated efficiently at the interfaces between an electron donor (D) and acceptor (A) materials in the bulk heterojunction (BHJ)7

(3)

68

architecture1, 3-5. The diffusion length of excitons in organic materials typically does not exceed 5-15 nm1, 3-5. Therefore, it is crucial to provide a large interfacial area between donor and acceptor materials, requiring maximal sizes of the phase-separated regions of ~10 nm.

Moreover, pathways for charge carriers to move towards electrodes are necessary for their extraction. These strong, but conflicting, requirements make morphology one of the key factors determining the overall power conversion efficiency of organic solar cells. So far, there is no solid theory to predict, or provide a systematic method to control, the self- organization of BHJ except for a few cases where general self-organization patterns were qualitatively computed8, 9. These challenges have driven the development of morphology characterization techniques.

Standard characterization methods such as electron or X-ray microscopy/spectroscopy are capable of achieving a spatial resolution below 10 nm with the help of contrast enhancement techniques2 such as energy filtering or selective staining of one of the materials with iodine vapors10. Sub-10 nm spatial resolution, however, is not easily achieved, in particular for the typically low contrast combinations of donor:acceptor materials used in organic photovoltaics. Though the list of available morphology characterization methods is fairly extensive2, morphology characterization methods have limitations, which depend on the particular method, on the particular types of donor:acceptor blends studied, and on the specific sample preparation methods2.

The major limiting factors in achieving a high spatial resolution in organic photovoltaic blends are usually the poor contrast between materials2, even when contrast enhancement techniques are used, and the sample degradation under exposure of the imaging X-ray or electron beams. Another drawback of many of the standard methods is that they cannot be applied to actual working devices; usually at least one of the electrodes has to be absent. This requirement poses a serious problem because the morphology typically changes by deposition or removal of the electrode11. A popular microscopy technique is Atomic Force Microscopy (AFM), which only provides information about the surface topography, a feature that is not necessarily representative for the bulk morphology. This technique is also not suitable for working devices, and the needed sub-10-nm resolution is problematic12.

Aiming to overcome these limitations, alternative methods were developed based on spectroscopic approaches, e.g. monitoring exciton diffusion by photoluminescence (PL) quenching13, 14 or by pump-probe spectroscopy15, 16. So far, these methods focused on exciton

(4)

69 generation in the polymer domain, which provides information on the polymer domain sizes.

Polymer excitons, however, are known to be delocalized over several repeating units17, 18, which limits the highest attainable spatial resolution. Additionally, the PL quenching method functions properly only if the PL quantum yield is reasonably high, which is incompatible with a high efficiency in organic solar cells.

Besides the strong absorption by the donor polymer, some fullerene derivatives, e.g.

PC71BM, show a substantial absorption of sunlight10, 19-21 as well. This opens a new opportunity to shed light on the size of the fullerene domains by studying the exciton diffusion in these domains, which can complement information obtained from other experiments on the length scale of the polymer domains.

In this Chapter, we demonstrate a new all-optical method for characterizing the nanoscale morphology of the fullerene domains in BHJs. The method is based on spectrally selective excitation of the soluble fullerene C70 derivative PC71BM, and detecting the time which is taken by the exciton to diffuse to the PC71BM:polymer interface where it splits into separated charges. The latter are detected by the charge-induced (polaron) absorption12 induced at the polymer backbone by the presence of positive charges (holes). For three polymers, regiorandom (RRa) and regioregular (RRe) P3HT as well as MDMO-PPV, which were selected as benchmark materials for exemplary cases of BHJ morphologies, we show that their blends contain charge-producing PC71BM clusters up to ~7 nm in size dispersed within the polymer matrix. The presence of larger PC71BM clusters is evidenced by the observation of a reduced efficiency of photon-to-charge harvesting. Unique spectroscopic signatures of subtle changes in BHJ morphology hold great promise for applications of the proposed technique for on-the-fly device characterisation of fully functional devices.

3.2. Results and discussion

3.2.1. Probing the morphology. Implementation

Spectrally selective excitation of the PC71BM followed by spectrally selective probing of the polymer allows spatial decoupling of the excitation and the probing processes.

Selective photoexcitation of PC71BM was achieved by tuning the excitation wavelength below the bandgap of the polymer were the PC71BM has a significantly higher absorption coefficient. Figure 3.1a-c shows the red edge of the absorption spectra of the three pristine polymers (dotted curves) and their blends with PC71BM (dashed curves). Based on the

(5)

70

highest contrast between PC71BM/polymer excitations (i.e. the ratios of their absorption spectra, solid lines), the excitation wavelength was selected as 680 nm for both RRa-P3HT and RRe-P3HT mixed with PC71BM, and 630 nm for PC71BM:MDMO-PPV.

Fig. 3.1 Absorption of PC71BM:polymer blends with three different polymers: (a) RRa-P3HT, (b) MDMO-PPV, (c) RRe-P3HT. The dotted and dashed lines depict absorption spectra of pristine polymers and their blends with PC71BM:polymer weight ratio 7:3, respectively. The solid lines present the ratio between optical densities of the blends and pristine polymers.

Exciton dissociation into charges was monitored by probing the charge-induced (polaron) absorption of the polymers in the mid-IR22-24. For this, the wavelength of the probe IR pulse was set close to the maximum of the low-energy polaron band at 3 μm for all three polymers (for polaron spectrum of RRa-P3HT, see Chapter 4). As the exciton is harvested (i.e. reaches the interface and dissociates into the charges), the charge-induced (polaron) absorption increases proportionally to the amount of holes in the polymer. By changing the delay between the pump and the probe pulses exciton diffusion followed by exciton dissociation, is monitored in the real time.

Dynamics of the exciton dissociation to charges are depicted in fig. 3.2 for different PC71BM/polymer weight ratios w. The transients were corrected for the pristine polymer response, present due to the finite contrast in excitation (this is especially important for the RRe-P3HT, fig. 3.1f). The correction was also applied for IR response of the PC71BM, noticeable for RRa-P3HT and MDMO-PPV at high values of w (see section S3.1 for more details). All transients were also normalized to the PC71BM absorption at the excitation wavelength (i.e. the transient amplitudes represent the charge yield per absorbed photon) to allow for direct comparison of the transient amplitudes at different PC71BM loads. The charge yields in fig. 3.2 was scaled by a constant number for each set of polymer:PC71BM blends separately in order to obtain the maximal amplitude of one for the low PC71BM load blends, which are expected to have a 100% exciton dissociation to charges efficiency. A more detailed explanation is presented in the section "Charge yield".

a) b) c)

(6)

71

Fig. 3.2 Charge yield as a function of time for PC71BM:polymer blends with different PC71BM weight fractions. Three types of polymers were used as donor materials: (a) RRa-P3HT (b) MDMO-PPV (c) RRe-P3HT. Symbols represent experimental data points after background subtraction; lines represent simulations. Irregularities in the transient amplitudes at low PC71BM content are due to low absorption over which the transients are normalized.

The charge yield dynamics for the blends with the three polymers have similar features that can be summarized in the following way: for the blends with low PC71BM content (low w) their respective transients exhibit a large amplitude and a rapid rise time (< 1 ps), whereas for the blends with high PC71BM content (high w) the amplitudes are decreased (except of the RRa-P3HT blends) while the rise of the response becomes substantially slower: up to 100 ps.

These dynamics are assigned to the PC71BM exciton diffusion followed by the dissociation to charges at the PC71BM:polymer interface. We attribute this behavior to variations in the PC71BM cluster size: the larger the PC71BM clusters, the longer it takes for excitons to reach an interface (therefore, the slower rise of the response) and more excitons are lost (therefore, the lower amplitude of the response).

a )

b )

c )

(7)

72

3.2.2. Data analysis 3.2.2.1. Charge yield

Fig. 3.3 shows the PC71BM exciton dissociation to charges yield obtained from the maximal amplitudes of charge-induced response (fig. 3.2). Blends with the low PC71BM/polymer weight ratio w, for instance, w ≤ 0.4, exhibit a high yield because PC71BM clusters are either small or not present at all (i.e. there are only isolated PC71BM molecules dispersed in the polymer). Taking into consideration ultrafast hole-transfer (HT) time (~30 fs)25 (see also Chapter 2), we can assume that most excitons, very close to 100%, dissociate into charges for the blends with low PC71BM content, e.g. w = 0.02. The charge yield (fig. 3.3) remains constant within experimental accuracy up to w = 0.4 for all PC71BM:polymer blends. Therefore, we assign the signal at low PC71BM content to the exciton-to-charges yield of unity, taking the average amplitude (shown in fig. S3.2) for blends with w = 0.02 ÷ 0.4 as the normalization factor.

Fig. 3.3 PC71BM charge yield calculated from the amplitudes of the pump-probe transients (fig. 3.2) for the blends of (a) PC71BM:RRa-P3HT, (b) PC71BM:MDMO-PPV and (c) PC71BM:RRe-P3HT. Symbols represent the experimental data, the lines present the results of the Monte-Carlo simulations. The red open symbols represent amplitudes at 1 ps, which are assigned to dissociation of the interface excitons, the green closed symbols are the maximal amplitudes, associated with all excitons that have reached interface, and the blue semi-closed symbols are the difference of the two, i.e. the excitons which originate from the bulk (i.e. outside the interface) of the PC71BM clusters.

b) a)

c)

(8)

73 Further increase of w results in a decrease of the overall charge yield due to the growth of PC71BM clusters, which prevents a significant number of excitons from reaching an interface. The dependence of the yield on w for the three polymers blended with the PC71BM is noticeably different. The RRa-P3HT blends (fig. 3.3a) do not exhibit clearly observable exciton losses even with the highest content of PC71BM (green closed symbols in fig. 3.3a).

High charge yields in the RRa-P3HT blends signify good intermixing between the PC71BM and the RRa-P3HT with mostly small PC71BM clusters forming up to w = 9. In contrast, the MDMO-PPV blends (fig. 3.3b) demonstrate a significant loss of exciton dissociation efficiency already at the PC71BM content of w~1.5. This is not surprising since MDMO-PPV blends with fullerene derivatives are known to form large fullerene domains above certain acceptor weight fraction (fig. S3.8), depending on the preparation method7, 25-28 (e.g. w = 1 to 4).

The PC71BM:RRe-P3HT blends (squares in fig. 3.3c) exhibit a decrease of the yield at w = 0.67 ÷ 1.5 similarly to the MDMO-PPV blends but to a significantly smaller extent. The observed difference of charge yield between the RRa-P3HT and the RRe-P3HT was expected because the morphology of the films that are formed by these two polymers is very different:

the film of RRa-P3HT is completely amorphous, but the film of RRe-P3HT forms semi- crystalline domains. The molecules of RRe-P3HT form nanocrystals prior to the aggregation of PC71BM during solution drying process29. Hence, most of the PC71BM molecules can only be dispersed in the disordered regions outside the nanocrystals of the RRe-P3HT film30. Consequently, PC71BM molecules have less volume to be dispersed within and, as a result, the PC71BM is pushed to aggregate into the clusters. Therefore, we assign exciton losses (fig.

3.3c) in the blends of RRe-P3HT with w = 0.67 ÷ 1.5 to the formation of the PC71BM cluster sizes comparable or larger than the exciton diffusion length.

Extrapolating this trend, one would expect the charge yield from the RRe-P3HT blends to further decrease above w = 1.5, however, the yield suddenly increases at w = 2.3. This unexpected and significant turn indicates an abrupt change in the nanostructure. The confirmation of this comes from the fact that linear absorption spectra demonstrate the disappearance of the red absorption shoulder of the RRe-P3HT near these blend compositions (fig. S2.2e in Chapter 2) associated with the absorption by the RRe-P3HT nanocrystals31. Others have also observed disruption of the RRe-P3HT nanocrystal at similar donor:acceptor compositions32, 33. Hence, the morphology of the RRe-P3HT blends becomes more similar to

(9)

74

that of the RRa-P3HT blends for w ≥ 2.3 because most of the nanocrystals are no longer present.

Summarizing the discussion above the three PC71BM:polymer blends exhibit very different charge yield as a function of the blend composition, which is reflected in the amplitude of the charge-induced response. Interestingly enough, the charge yield is sensitive to subtle changes in the morphology like the disappearance of nanocrystals in RRe-P3HT. To analyze the particular characteristic size of PC71BM domains, the dynamics of exciton diffusion should be considered.

3.2.2.2. Exciton dissociation represents their diffusion

The characteristic size of the PC71BM cluster can be estimated on the basis of excitons that dissociate via HT almost immediately after photoexcitation as they are generated at the interface with polymer (see also Chapter 2), and excitons, which are delayed because they are generated in the bulk of PC71BM and, therefore, have to diffuse prior to the dissociation.

These two very different time-scales, namely the HT (< 1 ps25) and the exciton diffusion (10- 100 ps20, 34), allow calculation of the fraction of excitons, which are generated at the interface with respect to the total number of excitons, by simply comparing the transient amplitudes at the respective times. This, in turn, provides a unique opportunity to make an estimate of the characteristic PC71BM cluster size as the ratio between the surface (interfacial) excitons and the entirety of (surface plus bulk) excitons through the surface-to-volume ratio. This ratio is inversely proportional to the linear size of the PC71BM cluster.

This idea is illustrated in fig. 3.3. The charge yield at 1 ps (red open symbols) and 100 ps (green closed symbols) delays is related to the interfacial and all harvested excitons, respectively. The difference between the two represents the bulk excitons (semi-open blue symbols in fig. 3.3). The general observation of all blends studied in this work is a decrease of the share of interfacial excitons when PC71BM content increases, which is assigned to a larger number of excitons harvested from the bulk of PC71BM clusters.

RRa-P3HT blends show the increasing percentage of the dissociated bulk excitons when w increases: the increase is steep for low w = 0.02 ÷ 0.43, while the increase becomes slower with larger w > 0.5. MDMO-PPV blends (triangles in fig. 3.3b) exhibit the steep increase of the bulk exciton fraction only in the range of w = 0.02 ÷ 0.05 when PC71BM content is low. These blends with further increase of PC71BM content w = 0.05 ÷ 1 show an almost constant share of dissociated bulk excitons while the portion of excitons created at the

(10)

75 interface as well as the total dissociation yield decrease. Fig. 3.3c shows the charge yield for RRe-P3HT blends (fig. 3.3c). The share of excitons created at the interface in these blends follows similar dependence on PC71BM load as the total yield of dissociated excitons. The dissociated bulk exciton fraction increases with the amount of PC71BM for w > 1 similarly to RRa-P3HT.

The RRe-P3HT blends, unlike the RRa-P3HT and the MDMO-PPV, do not show discernible dissociation of bulk excitons with w < 1.5 despite the fact that the overall charge yield decreases. When excitons are lost in the diffusion process, the appearance of extra charges is limited by the lifetime of excitons. Therefore, with the decrease of overall response the rise of the response limited by the finite (singlet) exciton lifetime (~500 ps in fig. S3.4, which is also consistent with fig. 4.6 in Chapter 4 and other reports20, 35) is also expected.

However, the experimental data presented in fig. 3.2 clearly exhibit charge-induced response saturation within the first 100 ps, which is much shorter than the exciton lifetime. This observation poses a question: why does the response saturate much earlier than exciton lifetime limitation?

This seemingly contradictory observation can be resolved in the following way. The majority of PC71BM clusters that produce charges are smaller than the exciton diffusion length so that the leveling-off of the experimental transients is limited by the cluster size but not the exciton lifetime. In turn, the loss of charge yield is caused by the PC71BM domains, which are so large that they contribute a negligible number of bulk excitons to the overall charge yield. Such a situation is only possible with a bimodal distribution of the characteristic PC71BM cluster sizes that are very different from each other. The bimodality can be explained by the concept of hierarchical morphology, present in the BHJ. The concept of hierarchical morphology is based on large phase separated fullerene domains that are surrounded by the polymer matrix, which contains dispersed fullerene molecules and small clusters30, 36, 37

. This argument is also applicable for the MDMO-PPV blends, which also exhibit a saturation of charge-induced response (charge yield) together with the decreasing overall amplitude. A more detailed analysis will be presented below.

3.2.3. Size of PC71BM aggregates 3.2.3.1. Monte Carlo simulations

To describe the dynamics of exciton diffusion in the PC71BM clusters, Monte Carlo (MC) simulations were used. The important advantage of the MC approach is that they allow

(11)

76

the inclusion of a non-flat energy landscape, which does not require finding an analytical solution of the diffusion equation.

Exciton diffusion was simulated as a random hopping between the discrete PC71BM molecules in a cluster of the close-packed spheres in the hexagonal arrangement; the cluster itself was modeled as a sphere. Initially, excitons are generated randomly in the PC71BM molecules within the cluster. Energetic disorder of the potential energy landscape of the PC71BM cluster was taken into account by the Gaussian disorder model38. Differences in energies of the PC71BM molecules were randomly assigned with the Gaussian distribution function with the standard deviation width of σ, a starting global fit value was 70 meV39. Exciton hopping rate kET between neighboring sites i and j with energies Ei and Ej, respectively, is described as a thermally activated process with the Boltzmann distribution:

(3.1)

where is the hopping rate when no thermal activation is required (exo-energetic/iso- energetic exciton hopping rate), is the Boltzmann constant, and is the temperature (fig.

3.4). The initial exciton hopping rate (no barrier) was kept as a global fit parameter with the starting value of 5 ps-1, which is consistent with the exciton diffusion length of ~5-6 nm40,

41 in the infinitely large PC71BM domain (fig. S3.4). Finally, exciton dissociation into charges at the surface of PC71BM cluster occurs with a finite hole transfer (HT) time25 (see also Chapter 2). The only variable parameter (per sample) associated with exciton diffusion was the number of PC71BM molecules, which was translated into PC71BM cluster size as the diameter of a sphere (fig. S3.6), whereas all other parameters, such as energy disorder and exo-energetic (or iso-energetic with a very small probability) exciton hopping rate, were kept global for all transients shown in fig. 3.2.

The loss of excitons is caused by the presence of large PC71BM domains, where most of the PC71BM excitons should not produce charges. However when the overall response becomes small then inputs from small and large PC71BM clusters may become comparable, which would result in an observable rise of the response at longer than 100 ps delays, therefore experimental data would not be reproduced adequately. The experimental data can be reproduced adequately only when the response (as small as it is) is dominated by the dissociation of excitons originating from small PC71BM clusters (a more detailed explanation

(12)

77 is presented in S3.2). Therefore, the simulations were run on each transient shown in fig. 3.2 with both, the small PC71BM clusters and with the large PC71BM domains, to make sure that the measured PC71BM domain sizes are large enough for simulations to reproduce the experimental data adequately. The average PC71BM domain size of each particular sample was measured with atomic force microscopy (AFM) (MDMO-PPV blends, fig. S3.8) and scanning transmission electron microscopy (STEM) (RRe-P3HT blends, fig. S3.9).

Fig. 3.4 Schematic representation of the Monte-Carlo simulation. (a) a schematic representation of the geometry of the simulation, (b) energy disorder model. Excitons are randomly generated within a PC71BM cluster (blue-red circle), diffuse by random hopping (pink arrows) to the interface where hole transfer takes place (green straight arrow). Each PC71BM molecule has a predefined random energy variation within energy disorder (b); the exciton hops from one molecule to another with the thermally- activated rate k0.

3.2.3.2. MC simulation results

The MC simulations were able to reproduce experimental data presented in fig. 3.2 (symbols) reasonably well. The simulation results shown in fig. 3.2 as solid lines demonstrate that the large PC71BM domains contribute negligible response (no exciton lifetime rise of the response is observed) and their surface indeed can be neglected, which means that these domains are spectroscopically invisible. The fact that the large PC71BM domains are spectroscopically invisible allows the detection of the volume fraction of large and small PC71BM clusters, which was used as a fitting parameter in order to reproduce the amplitudes of the transients in fig. 3.2 (see S3.2 for more details).

The results of the simulations are summarized in fig. 3.5 as solid lines – the PC71BM cluster sizes. The estimated PC71BM cluster size varies from 2 nm to 7 nm. The global fit parameters resulted in energy disorder meV (consistent with other report39) and exo-energetic exciton hopping rate ps-1. Both parameters represent a statistical average value of the whole ensemble. The energy disorder σ determines the time delay position when the response starts to saturate because the further arrival of excitons becomes

a) b)

(13)

78

much less probable when these excitons are trapped. The exo-energetic exciton hopping rate k0 determines how fast the excitons arrive at the interface, therefore, changing has a very similar effect to the change of the size of PC71BM cluster. The strong variation of does not make sense because PC71BM exciton diffusion length has to be realistic (~5-6 nm40, 41).

0 3 6

0 3 6

0.0 0.5 1.0 1.5 2 4 6 8 0

3 6

PC 71BM cluster size (nm) RRa-P3HT

MDMO-PPV

RRe-P3HT

PC71BM/polymer weight ratio (w)

Fig. 3.5 Estimated PC71BM cluster size dependence on the blend composition. The lines are results of Monte Carlo simulations while the symbols represent estimates made from surface volume / total volume ratio. (a) represents RRa-P3HT blends, (b) MDMO-PPV blends, (c) RRe-P3HT blends.

Summarizing the results, the PC71BM load can be ascribed to three main categories: 1) the low PC71BM load – w ≤ 0.11; 2) the average PC71BM load – w = 0.25 ÷ 1.5; 3) the high PC71BM load – w ≥ 2. The low PC71BM load blends of RRa-P3HT and MDMO-PPV exhibit the steep growth of small PC71BM clusters up to the size of ~3 nm. Similar PC71BM content dependence should be present in RRe-P3HT blends14, however, experimental data does not exhibit growth of small PC71BM clusters up to w = 1 due to limited experimental accuracy as the PC71BM excitation selectivity was too low (part of the absorption goes to the charge transfer state excitation as explained in Chapter 2). The growth of small PC71BM clusters with further PC71BM load appears to be rather slow in RRa-P3HT and MDMO-PPV blends as compared to low PC71BM loads. On the other hand, the growth of PC71BM clusters with increasing w of RRe-P3HT blends becomes comparable to that of RRa-P3HT. Note that the previously discussed disruption of nanocrystals of RRe-P3HT is reflected as the disappearance of the large PC71BM domains, shown in fig. 3.6a w = 1.5 ÷ 2.3.

The shortcoming of using the response from dissociated PC71BM excitons is that MC simulations cannot extract the particular size of large PC71BM domains where excitons do not reach the interface. The size of large PC71BM domains was obtained from the AFM for the

a)

b)

c)

(14)

79 MDMO-PPV blends (fig. S3.8) and from STEM for the RRe-P3HT blends (fig. S3.9); the results are shown in fig. 3.6b.

Fig. 3.6 Simulated volume fraction and size of large the PC71BM domains where most excitons do not reach the interface with a polymer. The top panel (a) represents the volume fraction while the bottom panel (b) shows the size of large PC71BM domains (obtained from AFM and STEM measurements).

The large (> 15 nm) PC71BM domains start to appear in the MDMO-PPV and the RRe-P3HT blends for w > 0.5. Coincidently, the small clusters of PC71BM do not exhibit significant growth with an increase of w. It may seem that with increasing PC71BM load not only the size and density of the large (> 15 nm) PC71BM domains should increase but also that the small (≤ 7 nm) PC71BM clusters should grow in size substantially, especially for the MDMO-PPV blends with high PC71BM loads. This unanticipated behavior can be explained in the following terms: when the PC71BM load increases, the density of small PC71BM clusters should increase; hence, these clusters would start to merge at some point forming large PC71BM domains. Consequently, the size of small PC71BM clusters would not increase substantially with the addition of extra PC71BM but the density of large PC71BM domains would definitely increase.

Another explanation for the slow growth of the small (≤ 7 nm) PC71BM clusters is that, in fact, these clusters are percolated with polymer chains or segments, therefore, effectively reducing their size. This effect is expected to be slightly more pronounced in the PC71BM:polymer blends as compared to the PC61BM:polymer because the PC71BM (unlike PC61BM) is a mixture of three isomers and hence, the PC71BM is expected to exhibit higher solubility and lower crystallinity, which consequently leads to higher spatial disorder.

Alternatively, it is possible that two PC71BM molecules might prefer to stick to each other a)

b)

(15)

80

entwining with their side chains already within a solution, before the drop-casted solution dries. Consequently, as PC71BM concentration in the solution would increase with an increase of PC71BM, the probability of PC71BM aggregation would also increase. This increasing probability of aggregation could account for the faster growth of PC71BM clusters up to w = 0.25. As soon as w would become larger than 0.25, the growth of the small PC71BM clusters would saturate, which could be caused by saturation of aggregation in the solutions used in this work (i.e. most PC71BM molecules are paired). Formation of larger PC71BM aggregates in the solution is not very likely. Therefore, further growth of PC71BM clusters would be driven by aggregation during solution drying process, which would not be as steep as the initial aggregation occurring already in the solution.

It cannot be entirely excluded that excitons are delocalized over two or more isoenergetic PC71BM molecules (i.e. PC71BM nanocrystal32) and dissociate immediately via hole transfer to the polymer. Therefore, the actual size of PC71BM clusters may be slightly larger than estimated. Referring to Kim et al.32 formation of PC61BM nanocrystals is not very pronounced with w ≤ 1 for MDMO-PPV and for RRa/RRe-P3HT blends. As was already indicated, the PC71BM is expected to be less likely to form nanocrystals as compared to PC61BM. Therefore, exciton delocalization should not play a significant role in underestimation of PC71BM cluster size at least for the relatively low PC71BM loads of w ≤ 1.

Moreover, due to the disappearance of the RRe-P3HT nanocrystals for w ≥ 1.5, RRe-P3HT becomes more amorphous, whereas PC71BM molecules are highly soluble in the amorphous regions of P3HT as compared to the nanocrystalline regions.

3.2.3.3. Short vs. long time-scale PC71BM exciton dissociation

Estimation of the PC71BM cluster size can be performed using a simple intuitive method: the calculation of the ratio of interfacial excitons to the total excitons. The interface/total exciton ratio is equal to the ratio of the volume of an outer PC71BM layer over the total volume:

, where R is the total radius of PC71BM cluster and r is the radius of inner sphere associated with the bulk excitons, VS and VT are surface volume and the total volume respectively, A(1 ps) and A(MAX) are charge yields at 1 ps and the maximum yield (~100 ps), respectively, and r=R-1 nm assuming the size of PC71BM molecule to be ~1 nm42. Using this simple approach, the diameter of the PC71BM domains was calculated for all blends (fig. 3.5 symbols). The PC71BM cluster size obtained using the

(16)

81 simple volume ratios is in excellent agreement with the MC simulations. This simple approach opens an opportunity to check quickly the characteristic size of PC71BM clusters on the fly even on the working devices.

3.3. Conclusions

An ultrafast pump-probe technique has been implemented to gain insights into the BHJ morphology of organic donor:acceptor materials for solar cells. This approach is complementary to the one demonstrated by Westenhoff et al.13 where the donor polymer was excited, as it allows estimating PC71BM cluster size. The proposed method is based on the selective photoexcitation of the acceptor material PC71BM and detecting the appearance of charges on the donor polymers, preceded by PC71BM exciton diffusion, by probing the low energy charge-induced (polaronic) transition at 0.4 eV. Unlike the excitation of polymer, which is delocalized along the polymer chain in the range of several nanometers43-45, excitation of PC71BM is usually localized in a single molecule. As the (de)localization length determines the minimum attainable spatial resolution, PC71BM excitation appears more advantageous than excitation of polymer domains as was done using Westenhoff et al.13 approach.

The amplitdue and time dependence of the charge yield dynamics have been shown to contain information on the PC71BM cluster size. A detailed modeling of the system was performed using MC simulations. The BHJ morphology of the organic photovoltaic blends studied here contains small PC71BM clusters of the size of several nanometers (up to 7 nm diameter) as well as the large PC71BM domains with sizes exceeding 15 nm, for which most excitons are lost. These observations are consistent with the paradigm of hierarchical morphology30, 36, 37.

Significant differences of BHJ morphology in terms of formation of the small (≤ 7 nm) PC71BM clusters and the large (> 15 nm) PC71BM domains have been observed for the mixtures with donor polymers of RRa-P3HT, MDMO-PPV, and RRe-P3HT. The small PC71BM clusters varied from 2 to 7 nm in PC71BM:RRa/RRe-P3HT blends and 2-4 nm in PC71BM:MDMO-PPV blends. MDMO-PPV blends exhibit a steeper increase of the size of the small PC71BM clusters than RRa-P3HT blends with an increase of PC71BM load.

RRa-P3HT did not show an appearance of large PC71BM domains within the whole range of PC71BM loads investigated. In contrast, the MDMO-PPV and RRe-P3HT based blends exhibit the formation of large PC71BM domains with PC71BM loads w > 0.4. An interesting

(17)

82

property of RRe-P3HT blends was observed: upon increasing the PC71BM load from w = 1.5 to w = 2 an abrupt increase of charge yield is observed, which was associated with the decrease of the volume fraction of the large PC71BM domains due to the disruption of the RRe-P3HT nanocrystals.

It has also been demonstrated that a simple rationale of taking the ratio between the interface PC71BM excitons, which dissociate within the first picosecond, and the total number of PC71BM excitons (e.g. all dissociated excitons when there are no losses) can reveal the size of PC71BM clusters. The outcome of this simple approach is in excellent agreement with the results of MC simulations.

The proposed method can provide a valuable feedback on how BHJ morphology should be optimized to balance between both the charge generation and the transport. The charge generation after excitation of PC71BM is especially important for modern solar cells involving narrow bandgap polymers, where PC71BM becomes the main absorber in the blue edge of the visible range. The limited number of excitons, which arrive at an interface is a direct consequence of both the PC71BM cluster size and the energy disorder. Therefore, optimization of BHJ morphology should involve not only tuning the size of PC71BM (or polymer) clusters, but also the ordering of the molecules, i.e. formation of nanocrystals.

3.4. Experimental methods

Poly[2-methoxy-5-(3′,7′-dimethyloctyloxy)-1,4-phenylenevinylene] (MDMO-PPV), regiorandom and regioregular poly(3-hexylthiophene-2,5-diyl) (P3HT) were purchased from Sigma-Aldrich. Regioregular P3HT has regioregularity greater than 90% head-to-tail regiospecific conformation. The soluble C70 derivative [6,6]-Phenyl C71 butyric acid methyl ester (a mixture of isomers19) (PC71BM) purity: >99% by HPLC with respect to the total fullerene content was purchased from Solenne.

Blends of MDMO-PPV and both P3HTs were prepared with different PC71BM content ranging from 0.02 to 9 as a PC71BM to polymer weight ratio w. The preparation procedure was the following: the polymer and PC71BM were dissolved separately with concentrations of 3 g/L for MDMO-PPV and 10 g/L for RRa/RRe-P3HT in 1,2-dichlorobenzene (ODCB) and stirred overnight on the hot plate with elevated temperature of 60˚. A solution of PC71BM was filtered using polytetrafluoroethylene (PTFE) filter with pore size of 0.2 μm. The two solutions of polymer and fullerene were mixed together with the appropriate volumes to

(18)

83 obtain the variety of PC71BM content in the solution. The final solutions were drop cast by equal volumes of 0.2 ml on the glass microscope cover slides with the thickness of 150 μm, and were allowed to dry. Evaporation of ODCB took, at least, several hours making the solvent-assisted annealing46, 47. During all the measurements, the samples were kept under the nitrogen atmosphere to prevent degradation; none was observed. Linear absorption was measured using standard Perkin Elmer spectrometer. Film thicknesses were measured using Dektak profilometer.

Time-resolved photo modulation spectroscopy was performed with a home-built setup.

The output of a 1 kHz Ti:Sapphire multipass amplifier was split to pump a noncollinear optical parametric amplifier (NOPA)48 and a 3-stages IR OPO49. NOPA was producing visible light output 30 fs, 40 μJ with wavelength tunability in the range of 500-700 nm. A 3- stage IR OPO49 was producing ~80 fs 350 cm-1 FWHM spectral width transform-limited pulses at 3.3 μm wavelength. The excitation wavelength was selected for the best absorption ratio between fullerene and polymer: for PC71BM:RRa/RRe-P3HT and PC71BM:MDMO-PPV it was 680 nm and 630 nm, respectively. The probe pulse was tuned to IR wavelength suitable for probing charge (polaron) appearance on MDMO-PPV24 and RRa-P3HT23 (fig. 4.7 in Chapter 4) as well as charge/Charge Transfer (CT) exciton in RRe-P3HT23, 50 at 3.3 μm. The system time resolution was ~100 fs. The visible pump was focused into a factor of 2 wider spot than the IR probe to minimize the spatial inhomogeneity of the pump.

The photoinduced absorption (PIA) response was calculated as the relative transmission change ΔT/T, where T – stands for transmission and ΔT – change in transmission. Pump flux was carefully tuned using gradient neutral density filter for the response to be in the linear regime. This resulted in pump fluxes of 75 μJ/cm2 for the P3HT blends and 120 μJ/cm2 for the MDMO-PPV blends. In all cases, the absorbed photon density was below 10-3 photons/nm3 (i.e. ~1 photon per 10 nm of length) to ensure a low probability of bi- exciton (non-geminate) annihilation.

The polarization of the probe beam, with respect to the pump, was rotated by 45˚ using the half-wave plate. The beam splitter was placed after the sample, splitting the IR probe beam into two. Two wire-grid polarizers (1:100 extinction), placed in the path of the two beams, were set to parallel and perpendicular directions with respect to the pump polarization. Two indium antimonide (InSb), liquid nitrogen cooled photodiodes were

(19)

84

simultaneously detecting two different polarizations of the signal. Parallel and perpendicular polarizations were used to recalculate isotropic component using the following relation51:

, (3.2)

where indices and denote parallel and perpendicular components, and t is a time delay.

The third InSb detector was used as a reference for IR pulses. The reference measurements together with a renormalization of digital signals greatly enhanced the signal-to-noise ratio of the PIA signal.

The precise pump-probe time-overlap position (zero delay) was carefully checked and if necessary, corrected before and after each scan (every 30 minutes) by measuring the reference sample. The reference sample was a blend of poly[2-methoxy-5-(2-ethylhexyloxy)- 1,4-phenylenevinylene] (MEH-PPV) mixed with 2,4,7-trinitrofluorenone (TNF) by weight ratio of 1:0.3. This blend forms a ground-state charge transfer complex for which response is limited only by apparatus resolution24, 50, 52, 53

. The materials were dissolved in chlorobenzene 2 g/L separately and mixed together. The final solution was drop-cast from chlorobenzene solution of 2 g/L on the same substrate as samples and allowed to dry. The root-mean-square drift of the reference zero delay was 5.5 fs during the 15-hour measuring session, which results in better than 5 fs accuracies in the zero position between the sample scans.

(20)

85

S3 Supplementary information

S3.1. Background subtraction

The background response of polymer and PC71BM was estimated by taking into account the absorption fractions of these two compounds. For the estimation of absorption fraction, the absorption coefficient was calculated using the following equation for each polymer:PC71BM blend and shown in fig. S3.1:

, (S3.1)

where OD – the optical density of a sample and L is the length of a sample in centimeters.

0 20 40 60 80

0 1000 2000 3000

(cm-1)

PC71BM fraction by weight (%)

y = 21 + 15 * x

at 680 nm

PC71BM:MDMO-PPV PC71BM:RRe-P3HT

PC71BM:RRa-P3HT

0 20 40 60 80

at 630 nm y = 117 + 29 * x y = 87 + 16 * x

PC71BM fraction by weight (%)

0 20 40 60 80

at 680 nm

PC71BM fraction by weight (%)

Fig. S3.1 Absorption coefficients (symbols) of (a) PC71BM:RRa-P3HT, (b) PC71BM:MDMO-PPV and (c) PC71BM:RRe-P3HT, calculated as optical densities divided by the film thickness at 680 nm (a and c) and 630 nm (b). The lines are the fits with the linear function (equations are given in the graphs) to the experimental data. The difference in the slope between (a), (c) and (b) is due to different wavelengths: at 630 nm, PC71BM absorbs more, by approximately a factor of 2 than it does at 680 nm.

The measured IR PIA response of PC71BM:polymer blends in this work originally had small (negligible in most cases) background response from pristine PC71BM and polymer materials (fig. S3.2). The shares of these responses depend on the relative absorption of the respective compounds. Therefore, the pristine polymer response (i.e. response from separated charges inside the polymer such as CT excitons) is expected to contribute more at low PC71BM loads while the response from pristine PC71BM becomes more visible at high PC71BM loads. For background subtraction, the responses of pristine PC71BM and pristine polymer, weighted by their relative absorption fractions, were calculated. fig. S3.2 (symbols) demonstrates the original data of photoinduced absorption -ΔT/T (isotropic component recalculated using eq. 3.1 presented in experimental methods of the main text) of the blends, where T and ΔT are the total transmission and change in transmission respectively. The share of the response of pristine PC71BM in the blends (fig. S3.2, solid blue lines) was estimated by rescaling the transient of pristine PC71BM film response (bottom panels) using the following equation:

a) b) c)

(21)

86

, (S3.2) where is the PIA response of pristine PC71BM film (fig. S3.2 bottom panel) with absorption A, ATotal is the total absorption of the particular blend and is the fraction of absorption by PC71BM, which is expressed as:

, (S3.3)

where the nominator represents the absorption by PC71BM and denominator represents absorption by PC71BM+polymer, x is the weight fraction of PC71BM from 0 to 1, the remaining weight fraction (1-x) is that of a polymer and x relates to the PC71BM/polymer weight ratio , k is the ratio between absorption coefficients of PC71BM and polymer.

The blends of PC71BM:RRa-P3HT, PC71BM:MDMO-PPV and PC71BM:RRe-P3HT exhibit k equal to 74, 25 and 18 respectively as determined from the linear fit (fig. 3.1).

To address the question of whether the PC71BM response is inherent for PC71BM molecules or it originates from PC71BM clusters, we performed separate measurements on PC71BM diluted in dichlorobenzene (not shown). These experiments did not reveal any response of an amplitude, that is, at least, 2-orders of magnitude lower than from the PC71BM film with comparable OD. Therefore, we can neglect the background response of isolated PC71BM molecules, which are probably mostly contained in the blends with small amounts of PC71BM (w = 0.02 ÷ 0.05). Therefore, the background correction for RRa-P3HT blends with low PC71BM loads was not performed. On the other hand, MDMO-PPV and RRe-P3HT blends with w < 0.42 had negligibly low PC71BM contribution (fig. S3.2) so that the background correction does not significantly change the PIA transients.

Similarly to PC71BM, the response of the pristine polymer in the blends (fig. S3.2 black dashed lines) was calculated by rescaling the transient of pristine polymer film response using the following equation:

, (S3.4) where is the photoinduced response dynamics of pristine polymer film (fig. S3.2 top panel) with the absorption A, and is the fraction of absorption by polymer expressed as:

(22)

87

. (S3.5)

Fig. S3.2 PIA transients with different weight ratios of PC71BM to polymer: (a) RRa-P3HT (b) MDMO-PPV (c) RRe-P3HT. Symbols represent measured dynamics of the PIA response. Solid blue lines and dashed black lines represent the PC71BM and the pristine polymer contributions, respectively, calculated from their relative absorption.

Accurate background subtraction relies on accurate measurements of the real absorption of the film. This is especially important for the films with low PC71BM content, where the absorption of a pristine polymer is extremely low and comparable to the surface reflection and scattering effects. RRa-P3HT and MDMO-PPV blends were assumed to reflect and scatter 15% of the incoming light referring to Lee et al.54.

In addition, for Fresnel losses, blends of pristine RRe-P3HT exhibited substantial light scattering, readily observable by the naked eye. An addition of extra PC71BM resulted in a noticeable change of the transmitted light, resembling the clear (i.e. without scattering) colored glass. The film scattering effects were verified using a laser pointer at the wavelength of 660 nm as a light source and a light detector placed just behind the sample to minimize the scattering effects. It was confirmed that dependence of optical density at 680 nm and 660 nm

a )

b )

c )

(23)

88

on w (fig. S3.3) is exactly the same (with a small offset in OD ~0.1 due to the different wavelength) except for pristine polymer and w = 0.02 blend. Therefore, scattering-corrected (laser measurement) values of RRe-P3HT were used when calculating the actual absorption of pristine RRe-P3HT and blend with w = 0.02.

0 20 40 60 80

0.0 0.2 0.4 0.6

OD-0.05 Laser = 660 nm

OD@680 nm

PC71BM fraction by weight (%)

OD Absorption spectrometer = 680 nm

Fig. S3.3 Linear absorption of PC71BM:RRe-P3HT as measured by a standard absorption spectrometer (full squares and a solid line) and a laser pointer operating at 660 nm (open squares and dotted line). The latter dependence was shifted by OD=0.1 to compensate for differences in absorption between 680 and 660 nm. Note that OD does not scale linearly with PC71BM weight because of different film thicknesses (for details, see fig. 2 in S2 of Chapter 2)

S3.2. Monte Carlo simulations

The Monte Carlo (MC) simulations require several parameters for modeling the data: 1) PC71BM exciton lifetime; 2) energy disorder of excitons on different PC71BM molecules (which results as energy differences Ei - Ej in Eq. 3.1); 3) initial (exo-energetic) exciton hop rate between PC71BM molecules (k0 in Eq. 3.1); 4) number of PC71BM molecules in the cluster sphere (PC71BM cluster size). Exciton lifetime was derived from photoluminescence measurements with streak camera (excitation wavelength 525 nm, apparatus function σ = 10 ps) shown in fig. S3.3. The estimated lifetime of 0.5 ns is comparable to the reported values20, 35. The starting energy disorder value of 70 meV was taken from reference39 as a global fit parameter. The starting, exo-energetic exciton hop rate was chosen 5 ps-1 to be consistent with the PC71BM exciton diffusion length of 5-6 nm40, 41 as shown in fig. S3.5 as a global fit parameter. fig. S3.5 demonstrates how energy disorder affects exciton diffusion: the increase of the distance traveled by the exciton slows down with time due to decreasing average exciton hopping rate. With the fixed exciton lifetime, the global energy disorder and

(24)

89 the global exo-energetic exciton hop rate in our simulations, the only variable parameter per sample remaining is the number of PC71BM molecules in the cluster.

0 200 400 600 800 1000 1200

0.1 1

Photoluminescence (arb. u.)

Delay (ps)

= 500 ps

Fig. S3.4 PC71BM photoluminescence measured with a streak camera. The symbols are the measured data. The line depicts the exponential fit with 500 ps decay.

1 10 100

1 2 3 4 5 6

Distance (nm)

Delay (ps)

0.01 0.1 1 10

Exciton hopping rate (ps-1 )

Fig. S3.5 Exciton hopping rate and diffusion distance dependencies on time after excitons are generated.

Energy disorder is 70 meV, initial hopping rate is 5 ps-1 and it gradually decreases due to loss of energy as excitons hop from one PC71BM molecule to another.

(25)

90

MC simulations have revealed that the simple approach of close-packed PC71BM molecules, approximated as spheres, was able to reproduce both amplitude and dynamics of all shown in fig. 3.2 with the cluster size shown in fig. 3.5. Naturally, simulation with only small (≤ 7 nm) PC71BM clusters was not sufficient to reproduce blends involving excitonic losses due to finite exciton lifetime (fig. S3.6a). Therefore, in addition to the small (≤ 7 nm) PC71BM clusters, which perfectly reproduce the exciton dissociation dynamics, we added the large (> 15 nm) PC71BM domains to account for excitonic losses. This bimodal PC71BM size distribution is called the hierarchical morphology29, 33, 34

. The approach of the hierarchical morphology is also in strong agreement with the estimated PC71BM cluster size by PC71BM surface excitons to the total excitons ratio (fig. 3.5).

0 25 50 75 100 125 150 175 200 0.0

0.5 1.0

 = 50

 = 36

PC71BM domain size

Normal distribution

Log-normal distribution 0.0

0.5 1.0

 = 15

 = 10

 = 50

 = 50 Normal distribution

Log-normal distr.

Probability (arb. u.)

 = 50

 = 30

Fig. S3.6 An example of simulations on PC71BM:MDMO-PPV blend with w=1.5. (a) the cluster size of 4 nm (dashed green line) reproduces the dynamics very well but totally misses the amplitude (the solid black line is obtained by dividing the amplitude of the dashed green line by a factor of 4.5). On the other hand, 15 nm cluster size (dotted blue line) does not reproduce the (initial) amplitude but fails in reproducing the dynamics. Normal (dashed green line) (b) and log-normal (solid black line) (c) distributions of PC71BM cluster sizes cannot reproduce the dynamics. (d) Shows the normal (red dashed lines) and log-normal (black solid lines) distribution of PC71BM domain sizes.

a )

b )

c)

a) b)

d)

(26)

91 We could assume that all excitons are simply lost in these large PC71BM domains

> 15 nm. Although most PC71BM excitons originating from large domains decay prior to reaching interface with a polymer, some of them are created close to the interface and therefore successfully dissociate. If the large (> 15 nm) PC71BM clusters are dominating in the blend this situation might lead to a very low but comparable response from large (> 15 nm) PC71BM domains (slow rise of the charge yield, limited by the exciton lifetime) and small (≤ 7 nm) PC71BM clusters (fast rise of the charge yield, not limited by exciton lifetime). Therefore, in order to make sure that exciton dissociation dynamics, which is not limited by exciton lifetime (the majority of charges are produced by small PC71BM clusters), is correctly modeled we involved the simulation of large PC71BM domains as well.

As simulations of the large PC71BM domains required long computational time because of the large number of molecules involved, the simulations were sped-up by simplifying a large PC71BM cluster to uniform sphere representing a PC71BM domain, where excitons are randomly generated within a sphere at any location and have freedom to move to any direction a distance of 1 nm. The simplification is possible because the difference between close-packed spheres of PC71BM molecules and continuous spheres is negligible for a domain size above 15 nm diameter (fig. S3.7). The close-packed sphere model is based on adding PC71BM molecules in such a way that the distance to the center of the cluster would be minimized. The size of PC71BM clusters composed of molecules was calculated as the distance from the last PC71BM molecule, added to the simulation, to the center of the cluster (fig. S3.7 red circles). Note that the size of PC71BM cluster with a small number of molecules (e.g. < 4) is slightly overestimated due to the imperfect shape of the cluster (fig. S3.7 red circles): size of the cluster consisting of 2-3 molecules has the length scale of exciton diffusion of 2 nm in some directions and 1 nm in the other directions.

The equivalent diameter of the uniform sphere (fig. S3.7 blue triangles) was recalculated from the following relation:

(S3.6)

where R is the radius of an equivalent uniform sphere, r = 0.5 nm is the approximated radius of PC71BM molecule and n is the number of PC71BM molecules. Naturally, the two methods of estimation of PC71BM domain size have some mismatch caused by the fact that a uniform sphere assumes the complete filling of the three-dimensional space within a sphere, whereas

Referenties

GERELATEERDE DOCUMENTEN

After concluding the first chapter by a literature review about regional competitiveness, in Chapter 2, Paris (Ile de France) will be taken under focus as it is the second leading

By means of suitable individual extensions of basis elements, we construct a linear continuous exten- sion operator, when it exists for the corresponding space.. In general, elements

Onze lee boekjes Joe en Roe en 0 n Ei gen Boe k met de daarbIj behorende taalb ekje zijn ont taan tijd ns de bewerking van het ede r·.. land Taalbo k

Om op d ze onh rbergzame 3.&lt;'l.rd Het groenend wier gelijk te zijn, I al de rijkdom, dien ik mij.. 'it' goedertiern n Vaders hand, Op zijn genade

A second approach to amplify motion is via dynamic macromolecules with the per- spective to design responsive and mechanical polymer mate- rials that is, fibers, networks, gels

Both of these examples – one active and one passive – would benefit from the (nanoscale) objects of interest being elevated from the surface and held in the center of the

Furthermore, it reflects on the integration of both optical and electrochemical detection of oxygen into microfluidic devices for control of oxygen gradients during cell and

Met geen blik verried hij zijn gedachten, en hoe hij naar het alleenzijn met haar verlangde. De dag duurde eindeloos lang. 's Middags maakte hij, geheel in tegen- stelling met