Stabilizing polymer-based bulk heterojunction solar cells via crosslinking

(1)

1346

Received: 5 October 2013 Revised: 18 December 2013 Accepted article published: 24 January 2014 Published online in Wiley Online Library: 26 February 2014

(wileyonlinelibrary.com) DOI 10.1002/pi.4712

Stabilizing polymer-based bulk heterojunction solar cells via crosslinking

Guillaume Wantz,

^a*

Lionel Derue,

^a

Olivier Dautel,

^b

Agnès Rivaton,

^c

Piétrick Hudhomme

^d

and Christine Dagron-Lartigau

^e*

Abstract

The active layer of a polymer photovoltaic cell is mainly based on a blend of two components: a semiconducting polymer (electron donor) and a fullerene derivative (electron acceptor) to form the bulk heterojunction (BHJ). To offer optimum photovoltaic performances, the morphology of this layer has to be very carefully controlled at the nanoscale. The materials of the BHJ require specific phase segregation enabling the optimum photogenerated exciton diffusion and dissociation, and also to ensure pathways for charge carriers to electrodes. However, such a specific morphology is thermodynamically unstable over time and phase segregation occurs with thermal cycling under solar operating conditions inducing a decrease of solar cell efficiency. This review reports on the recent progress towards obtaining a stable optimized BHJ morphology and improved efficiency stability, using different chemical routes for crosslinking the organic semiconductors.

Keywords:photovoltaic; polymer; fullerene; crosslink; morphology; stability

INTRODUCTION

Organic photovoltaic (OPV) cells have attracted a tremendous amount of interest during the last decade. Being able to produce low-cost, large-area, printed OPV modules is indeed a major chal- lenge to answer renewable energy needs. In order to commercial- ize this technology, three conditions have to be met: a satisfactory power conversion efficiency (PCE); a proven cost-effective method of manufacturing; and a long-enough lifetime of modules under operating conditions. Around a thousand scientific publications are currently published every year on OPVs. Most of them deal with the development of new materials and new device architectures to improve the PCE as the main work-horse of the community. While the highest PCE of 9.2% was described in 2012,¹the stability of devices is not ensured yet and is still a subject of research that needs to be further improved in the future.

It is now recognized that the stability of OPV cells suffers from three main failure mechanisms even with a high-performing encapsulation. Firstly, there are multiple interfacial issues since these devices are made of stacks of layers to be carefully controlled at the nanometre scale. Electrodes can be sensitive to oxidation and interlayer atoms can also diffuse into the active layers. For example, we recently demonstrated that atoms of Mo, O and Ag can diffuse from the top layers of MoO₃/Ag of inverted solar cells into the active layer upon thermal stresses leading to reduced performances.²Secondly, organic polymer semiconductors suffer from photodegradation under exposure to sun- light. Indeed it is well known that one major problem associated with the application of polymers in general is their instability to weathering.³^,⁴The photoactive materials used in polymer solar cells are conjugated polymers, in which 𝜋-electrons are highly delocalized with consequent electro-optical properties such as solar light absorption. The direct absorption of UV–visible light by the chromophores of the polymer matrix may be responsible

for direct photochemical reactions. It has been evidenced that, under the impact of light, aromatic polymers can undergo various reactions without involvement of oxygen, such as rearrange- ment, chain scissions or crosslinking.⁵Those reactions provoke the breaking of the conjugated backbone, resulting in a loss of 𝜋-electron conjugation.⁶^,⁷ As a result of polymer photobleach- ing, the device performances of organic solar cells are reduced.

Finally, the morphology of the polymer–fullerene bulk heterojunction (BHJ) composing the active layer appears to be not stable thermodynamically. To achieve eﬃcient devices, the phase segregation between these materials has to be tuned to form a bi-continuous network with domain sizes of the order of magni- tude of the exciton diﬀusion lengths (below 20–30 nm). However,

∗ Correspondence to: Guillaume Wantz, IMS, Université de Bordeaux, UMR CNRS 5218, ENSCBP, 16 Avenue Pey Berland, F-33607, Pessac, France.

E-mail: guillaume.wantz@ims-bordeaux.fr and Christine Dagron-Lartigau, IPREM CNRS-UMR 5254, Université de Pau et des Pays de l’Adour, Hélioparc, 2 avenue Président Angot, 64053, Pau Cedex 9, France. E-mail:

christine.dagron-lartigau@univ-pau.fr

a IMS, Université de Bordeaux, UMR CNRS 5218, ENSCBP, 16 Avenue Pey Berland, F-33607 Pessac, France

b AM2N, UMR 5253, Institut Charles Gerhardt, 8 rue de l’Ecole Normale, 34296, Montpellier, France

c ICCF, Université Blaise Pascal, F-63000 Clermont-Ferrand, France

d MOLTECH-Anjou, Université d’Angers, UMR CNRS 6200, 2 Boulevard Lavoisier, 49045 Angers, France

e IPREM CNRS-UMR 5254, Université de Pau et des Pays de l’Adour, Hélioparc, 2 avenue Président Angot, 64053 Pau Cedex 9, France

(2)

1347

Dr Guillaume Wantzis associate professor in the Laboratory IMS (UMR CNRS 5218) at the University of Bordeaux (France) since 2006. His research interest is in organic electronics with a focus on polymer photovoltaic solar cells and light-emitting electrochemi- cal cells. His research deals with device fabrication and optimization through careful control of nanoscale morpholo-

gies and the development of original device architectures.

He is currently developing strategies to improve the stability of polymer bulk heterojunctions for long-lived solar cells.

Dr Lionel Deruereceived his BS degree (2008) and MS degree (2010) from the Department of Physics at the Univer- sity of Mons (Belgium) with a the- sis on organic light-emitting diodes.

He obtained his PhD at the Univer- sity of Bordeaux in 2013. From Jan- uary 2014, he is a postdoctoral fel- low at the Ecole Polytechnique (Paris).

His research interests include device

engineering and device physics in organic optoelectronics.

Dr Olivier Dautelis CNRS researcher in the Charles Gerhardt Institute (UMR CNRS 5253) at the Graduate School of Chemistry of Montpellier (France). He works on the design and the synthesis of 𝜋-conjugated molecules for opto- electronic applications. His research is on organic electronics with a focus on the control of supramolecular organization of molecular assem-

blies in order to control their optoelectronic properties.

such a morphology with nanosized crystallites of fullerene is not stable over time under thermal stress and/or solar cell operating conditions. This morphology tends to segregate and microscale crystals of fullerene derivative appear after ageing under opera- tion. The driving force is either the self-organization of crystalline polymers expelling fullerenes out of their domains or Ostwald ripening.⁸Figure 1 shows for example optical microscope images of poly(3-hexylthiophene) (P3HT):[6,6]-phenyl-C61-butyric acid methyl ester (PCBM) layers before and after being thermally annealed at 150∘C, as the most widely cited polymer–fullerene system in the literature.⁹After 5 h at 150∘C, very large fullerene crystals have grown, leading to the creation of PCBM-depleted areas (purple in the ﬁgure). In this scenario, the orange-looking domain is still active for the photovoltaic process but the purple areas are no longer producing electricity. After 24 h at 150 ∘C, the P3HT:PCBM blend is entirely covered with large crystallized domains of fullerenes resulting in the disappearance of an appropriate BHJ and complete loss of solar cell performance. Similar crystal growth has been reported many times by multiple groups for poly(phenylene vinylene) (PPV) derivatives,¹⁰ P3HT¹¹^,¹² or other low-bandgap state-of-the-art polymers.¹³

Dr Agnès Rivaton is Director of Research at CNRS in the photochem- istry team of the Institute of Chemistry of Clermont-Ferrand (France). She develops a physicochemical approach to the photochemical behaviour of polymers from mechanisms of photodegradation to properties of materials. One important area of application is energy sources,

with projects focused on the stability of organic solar cells.

Dr Piétrick Hudhomme is professor in the Laboratory MOLTECH-Anjou at the University of Angers (France).

His research interests concern the synthesis of functional photo- and/or electroactive molecules for applications in supramolecular chemistry, molecular electronics and organic photovoltaics. In this last topic, he develops new acceptor materials, especially in fullerene and perylenediimide series, with the aim

of improving stability and absorption of organic solar cells.

Dr Christine Dagron-Lartigauis assis- tant professor at the University of Pau in the Institute IPREM (France). She works on the design and synthesis of p-type conjugated (co)polymers for organic solar cells. She develops various structures and architectures to improve their properties, absorption and stability, and to provide materials that ensure morphological stability.

The main topic of this review concerns the specific point of how to stabilize the BHJ at the nanoscale. There are multiple techniques under development to stabilize the BHJ and con- sequently improve performance stability: the use of ternary blends with compatibilizers such as block copolymers,¹⁴–¹⁶ amorphous fullerene derivatives,¹⁷ high glass transition temperature polymers,¹⁸ thermocleavable polymers,¹⁹ specifically designed donor–acceptor systems with enhanced interactions,²⁰ functionalized side-chains on the polymer²¹ or crosslinkable materials. This mini-review focuses specifically on the crosslinking approaches for OPV cells, since a state of the art already exists for organic light-emitting diodes.²²During the last decade there have been numerous investigations of the use of crosslinkable materials in OPV cells. Crosslinking consists of branching molecules one to the other by covalent bonding using specific materials or various stimuli (temperature, UV light, photoinitiator, etc.), and is promoted in the solid-state BHJ during device fabrication to freeze the morphology in its optimized form.

Various crosslinking approaches can be considered since the BHJ is composed of two materials, an electron-donating polymer (D) and electron acceptor (A). One can crosslink the donor to the donor (D–D), the donor to the acceptor (D–A) or the

(3)

1348

C₆H₁₃

S

O O n P3HT

PC₆₁BM

Figure 1.Optical microscope images of P3HT:PCBM layers before and after being thermally annealed at 150∘C for 5 or 24 h (Derue L, Lebail M, Hirsch L and Wantz G, unpublished).

Figure 2.Scheme showing the BHJ (top) and three crosslinking pathways (bottom). The red line represents the bridge as a result of crosslinking chemistry between donor and donor, between acceptor and donor and between acceptor and acceptor.

acceptor to the acceptor (A–A). These three approaches are illustrated in Fig. 2 and are successively developed in this review.

Among the various crosslinking technologies developed so far, Fig. 3 shows the functions that have already been employed in organic electronics and more specifically in OPVs. The choice of functional group has to be made in accordance with the crosslinking external stimulus that should be compatible with the conditions of device fabrication. For example, the use of sol–gel chemistry implies activation by the presence of acid, base, fluoride²³ or the activation of photogenerated acids by UV light stress.²⁴ Epoxide and oxetane groups are also known to require a photoinitiator or acid initiator for further thermal or photo-crosslinking. Some other crosslinking functions are activated by UV light at various wavelengths, such as bro- mide, azide, anthracene or cinnamate. Azide groups can also promote crosslinking reactions upon thermal annealing, as well as alkenyl, diethynyl or styrene groups. In any case, the crosslinking stimulus must be compatible with the solar cell fabrication process without affecting or degrading other layers. Furthermore, it is necessary to develop techniques that will be easily transferable to large-scale, large-area manufacturing.

CROSSLINKING DONOR TO DONOR (D–D APPROACH)

In this approach, the crosslinkable functions are generally attached on the lateral chains of a conjugated backbone. The modiﬁed monomer can be used in a homopolymer, but is generally used

as a comonomer to randomly disperse the crosslinkable functions along the conjugated backbone, given that a few percent are suﬃ- cient to create an insoluble network while retaining the structura- tion of the polymer. The crosslinked network may thus hinder the diﬀusion of PCBM in the BHJ. The polymers described in the frame- work of this approach are presented in Fig. 4.

One of the ﬁrst demonstrations of an eﬃcient crosslink reducing the formation of PCBM crystals (after thermal annealing at 150

∘C for 2 h; Fig. 5) was described by Miyanishiet al.²⁵ The stabilization of the BHJ during annealing time was achieved using poly[3-(5-hexenyl)thiophene] (P3HNT) in which the crosslinkable function was an alkenyl group (Fig. 4). The PCE of these devices showed a smaller decrease (from 3.03 to 1.74%) than for non-crosslinked P3HT (3.11 to 1.00%) after 10 h at 150∘C, even if the formation of PCBM crystals was not completely avoided.

A strategy using bromine-functionalized P3HT (P3HT-Br; Fig. 4) copolymers crosslinked using UV treatment at 254 nm was reported by Kim et al.²⁶ Devices prepared with the crosslinked P3HT-Br in blends with PCBM exhibited an absence of PCBM crystals (after 24 h at 150∘C) and an improved PCE stability, even after 2 days at 150∘C, due to the stabilization of the BHJ (Fig. 6).

High stability with an average PCE of 2.2% was also observed for bilayer devices fabricated with P3HT-Br10:PCBM.

This Br-involving method was also applied by the same team to a donor–acceptor copolymer (low-bandgap polymer) based on N-alkylthieno[3,4-c]pyrrole-4,6-dione (Fig. 4).²⁷ The crosslink was optimized for an amount of 16% mol of bromine for this comonomer using UV treatment (254 nm), but was not possible by thermal annealing. While the non-crosslinked BHJ exhibited

(4)

1349

Figure 3.Crosslinkable chemical functions reported in the literature concerning organic semiconductors. Below each chemical structure the references are mentioned as well as the name. The related crosslinking stimuli that can be used to initiate the chemistry are also given.

a decrease of PCE from 5.6 to 4.0 % after 72 h at 150∘C, the crosslinked BHJ showed an increase from 3.3 to 4.6% for the same ageing treatment, yielding the highest PCE obtained so far with crosslinked p-type polymers. Very recently analogues of poly(cyclopentadithiophene-alt-benzothiadiazole) (Fig. 4), with an alkenyl group modiﬁed or not by anisole or bromobenzene, were used by Waterset al.in solar cells.²⁸The crosslink was promoted by high-temperature annealing (260 ∘C), which could be a problem in some cases due to possible degradation of the poly(3,4-ethylenedioxythiophene) (PEDOT):polystyrenesulfonate (PSS) layer and the impossibility of using polymer substrates, but allowed an improvement of hole mobility. The initial PCEs obtained for non-crosslinked and crosslinked polymers were nearly identical (3.88 and 3.65%, respectively), but the half-life was increased by 45% in the latter case (for encapsulated devices illuminated at AM1.5).

Another important class of crosslinkable function is repre- sented by azide groups. The crosslinking of such functionalized polythiophenes was achieved with a high-temperature treatment (above 185∘C) and was ﬁrst demonstrated by Murrayet al.

in 1996.²⁹^,³⁰ Azides are versatile functions that can promote crosslinks between donor polymer chains upon UV irradiation or promote cycloaddition reaction with PCBM with a thermal treatment at 150∘C³¹(more details are given in the next section) (Fig. 7). In that study, UV irradiation allowed the ﬁrst demonstra- tion of a solvent-resistant organic transistor, without disrupting the molecular ordering and charge transport. The thermal treatment allowed the stabilization of the OPV BHJ with PCBM, even after 40 h annealing at 150∘C. Namet al.described another use of P3HT-N₃to crosslink the BHJ using UV light, still with improved thermal stability of OPV performance.³²

Crosslinked poly(ﬂuorene-alt-bithiophene) conjugated polymers bearing oxetane functions were prepared by Charaset al.

and were activated with a photo-acid generator in the presence of UV light (Fig. 4). This allows patterning for various organic electronic applications.³³ Using this technique, nanostructured donor–acceptor interfaces with the formation of columnar-grain ﬁlms were prepared for OPV cells using a similar polymer, F8T2Ox1 (Fig. 4).³⁴The oxetane strategy has also been recently successfully applied by the same team to a P3HT-based polymer (P3HT-Ox10;

(5)

1350

Figure 4. Examples of some functionalized side-chain semiconducting polymers: poly[3-(5-hexenyl)thiophene] (P3HNT); bromine-functionalized P3HT (P3HT-Br); azide-functionalized P3HT (P3HT-N₃); oxetane-functionalized polymer; bromine-functionalizedN-alkylthieno[3,4-c]pyrrole-4,6-dione (TPD-Br); poly(cyclopentadithiophene-alt-benzothiadiazole) (Cr-PCPDTBT); F8T2Ox1; P3HT-Ox10; PTcbpTT; and bromine-, azide- and vinyl-functionalized poly(benzo[1,2-b:4,5-b^′]dithiophene-alt-thieno[3,4-b]-thiophene) (PBT).

(6)

1351

Figure 5.Optical microscope images of (a) P3HT:PCBM and (b) P3HNT:PCBM ﬁlms after thermal annealing at 150∘C. Annealing times are shown in the images (scale bar: 200μm).²⁵(Reprinted (adapted) with permission from Miyanishiet al.²⁵Copyright 2009 American Chemical Society.)

(a) (b)

Figure 6.Photo-crosslinking behaviour of P3HT-Br copolymers. (a) Insoluble film fraction as a function of UV exposure time measured for three different P3HT-Br copolymers having 5, 10 and 20 mol% Br units as indicated. (b) Efficiencies of devices containing P3HT:PCBM and P3HT-Br10:PCBM annealed at 150∘C for different times, without and with UV treatment for 10 and 30 min.²⁶(Reproduced with permission from Kimet al.²⁶)

Fig. 4).³⁵A recent study by Carléet al.reports a comparative use of bromine-, azide-, alkenyl- and oxetane-functionalized polymers for solar cells.³⁶In all cases, more or less insoluble polymers were obtained evidencing the efficiency of crosslinking. However, that study does not provide any absolute PCE value, so stability curves plotted with normalized PCE could be speculative depending on the initial PCE achieved with each of these compounds. A similar study was very recently reported by Chen et al. with the aim of comparing vinyl, bromine and azide photo-crosslinkable groups grafted on the low-bandgap conjugated polymer poly(benzo[1,2-b:4,5-b^′]dithiophene-alt-thieno[3,4-b]-thiophene) (PBT; Fig. 4).³⁷ The initial PCEs are very close (between 2.5 and 2.7%) and then decrease differently depending on the crosslinking group. Probably due to different crosslinking mechanisms (induced by UV light at 254 nm), it was shown that solar cells based on PBT-Br and PBT-N3 exhibited more thermally stable device performance (after annealing at 150∘C for 12 h) by main- taining 78 and 66% of their initial PCE values, respectively, whereas PBT-vinyl-based device retained only 51% of its initial PCE value.

These results are in accordance with the morphology observed in optical microscope images of thin films from blends of different PBT-based polymers with PCBM (Fig. 8), showing that bromine is the most efficient crosslinking group to limit the formation of PCBM crystals.

In the various examples presented in this section, crosslinked polymers retain molecular packing and good mobilities, without degrading optical and electrical properties, and allow a certain stabilization of OPV performance upon thermal annealing in the case of bromine and azide compared to non-crosslinked material.

Bromine groups were also applied to maintain the ordered nanostructure of liquid-crystalline conjugated materials (polythiophene functionalized with mesogenic cyanobiphenyl) with UV treatment, as reported by Liet al.³⁸ The stabilization by crosslinking improved the obtained PCE from 0.5% (non-crosslinked) to 1.2%.

Another example of crosslink chemistry was shown by Ouhibet al.

using diblock copolymers functionalized with acrylates to improve PCE stability with annealing time.³⁹Solar cells prepared with the crosslinked copolymer retained more than 85% of initial PCE after

(7)

1352

Figure 7.Left: possible reaction routes depending on the stimulus used: crosslink between P3HT chains with UV light or cycloaddition with PCBM with heat treatment at 150∘C.³¹Top-left image: dissolution of the P3HT layer when not crosslinked; top-right image: insoluble ﬁlm made of crosslinked P3HT functionalized with azide. Bottom-left and bottom-right images: optical microscope images showing the eﬀect of adding 0 and 15% of P3HT-azide in the P3HT/PCBM blend on the formation of PCBM crystals in the active layer of OPV devices. (Reprinted (adapted) with permission from Kimet al.³¹Copyright 2012 American Chemical Society.)

165 h at 110∘C compared to 65% for P3HT under the same conditions. However, in this case, the crosslinked network of polythiophene did not fully prevent the formation of PCBM crystals (after 40 h at 110∘C).

With this approach, it is possible to use various functions to promote crosslinking reactions. Some of them can be promoted under mild conditions, with thermal or UV treatment. Moreover, some of them such as bromine or vinyl groups can be introduced on lateral chains without markedly aﬀecting the molecular packing of polythiophenes. Copolymerization of usual monomers with comonomers bearing crosslinking functions allows control of crosslinking moieties that is critical to achieve optimal performances. However, while the donor–donor crosslink approach

improves performance stability over time, it does not completely hinder the formation of PCBM crystals.

CROSSLINKING DONOR TO ACCEPTOR (D–A APPROACH)

With the objective of preventing the formation of micromet- ric PCBM crystals, various strategies have been investigated. As already mentioned, the formation of a donor–donor reticulated network slows fullerene diﬀusion in the active layer resulting in an increase of thermal stability of optoelectronic devices. However, a completely frozen morphology cannot be achieved yet. In this

Figure 8.Optical microscope images of spin-coated thin ﬁlms of PBT, PBT-vinyl, PBT-Br and PBT-N₃polymers blended with PCBM (1:1 wt) (a–d) before and (e–h) after annealing at 150∘C for 12 h. All the thin ﬁlms were irradiated with UV light at 254 nm for 30 min before the annealing process. Scale bar:

100μm.³⁷(Reproduced with permission from Chenet al.³⁷)

(8)

1353

context, it appears that linking the acceptor to itself (A–A) or to the donor polymer (D–A) could be more eﬀective against macrophase separation. In this case, the links (A–A or D–A) can be realized after deposition and optimization of the BHJ.

As mentioned above, Kim et al.³¹ have developed an original approach to achieve thermally stable organic electronic devices by introducing an azide crosslinkable group into a polythiophene backbone. The azide function offers the possibility to crosslink following two chemical pathways (Fig. 7) depending on the stimuli used. On the one hand, the azide could be photochemically decomposed into a molecule of nitrogen and a highly reactive nitrene function. In this case, the generated nitrene radical could react without any selectivity with the donor polymer (D–D crosslinking) and with the acceptor (fullerene derivative, D–A crosslinking). On the other hand, the azide function could thermally (at low temperature) and selectively react with the fullerene derivative through 1,3-dipolar cycloaddition to give triazolino(4^′,5^′:1,2)fullerenes and subsequent transfor- mation at higher temperature into the respective azafulleroids and/or aziridinofullerenes with extrusion of nitrogen molecules (D–A crosslinking).⁴⁰ In the latter case, no reaction could occur between the azide backbone and the donor backbone. The concept was demonstrated through the synthesis of polythiophenes bearing azide groups attached to the end of the alkyl chain (P3HT-azide). Even if azide-functionalized P3HT derivatives were first described by Murray et al.,²⁹ Kimet al.³¹ developed a straightforward approach by reaction of a halogenated side chain with sodium azide using 18-crown-6. Thermal treatment at 150∘C of P3HT devices containing 15% of P3HT-azide10 copolymer with PCBM as the electron acceptor resulted in the reaction of the azide group with the fullerene. As a consequence, the BHJ of solar cells containing 15% of P3HT-azide copolymers showed an average efficiency higher than 3.3% after 40 h annealing at 150∘C. This enhanced stability is due to a compatibilizer that formsin situat the P3HT–PCBM interface and suppresses macrophase separation (Fig. 9).

Gholamkhass and Holdcroft⁴¹developed new copolymers based on P3HT and polystyrene blocks using azide functions introduced on the polystyrene block (Fig. 10). BHJ devices were fabricated from azide-functionalized graft copolymer possessing a graft-chain density of 1% (n=0.01) and PCBM. As in the previ- ous example, a thermal stimulus helps trigger a solid-state reaction between the azide groups attached to the copolymer and PCBM.

As a consequence, ﬁlms were insoluble and the morphology was stabilized, as evidenced by limited growth of macroscopic crystals of PCBM over time (Fig. 11(a)). Photovoltaic devices prepared from these stabilized layers exhibited 1.85% PCE which dropped to 0.93% over 3 h at 150 ∘C, whereas PCE of regular P3HT/PCBM devices, initially 2.5%, dropped to 0.5% over the same period (Figs 11(b) and (c))

By functionalizing a conjugated polymer, still at the end-functionalization of the polymeric chains, Khiev et al.⁴² recently reported an unprecedented approach. Indeed, the use of P3HT end-capped with anthracene (P3HT-A) in a blend with PCBM was demonstrated to morphologically stabilize BHJ solar cells. Using a simple UV-curing step, the P3HT-A chain reacted with PCBM fullerene via a [2+2] cycloaddition and stabilized the blend (Fig. 12). Optical microscopy images showed a decrease of the formation of PCBM crystals as a result of using P3HT-A when UV treatment (365 nm) was performed before thermal annealing. Even if the eﬀects of UV treatment durations on the current–voltage curves were diﬃcult to explain, an important

result stands on the fact that the device performances obtained were not reduced after diﬀerent UV treatments.

The D–A strategy was principally addressed according to the functionalization of P3HT with crosslinkable functions able to react with the PCBM.³¹^,⁴¹^,⁴²In this case, the chemistry is straightforward since only the polymer needs to be synthesized. However, very few functions are able to react speciﬁcally with PCBM. In particular, concerted additions of an unsaturated system (azide or diene) through [1,3] dipolar or [4+2] cycloaddition to the fullerene could only be envisaged. Nevertheless, this approach may be applicable to all types of fullerene derivatives (PC60BM, PC70BM, indene-C60

bisadduct (IC₆₀BA), etc.).

CROSSLINKING ACCEPTOR TO ACCEPTOR (A–A APPROACH)

The stabilization of the BHJ nanomorphology could also be achieved by realizing a crosslinking process from a conveniently functionalized acceptor fullerene derivative.⁴³ With the aim of reducing phase separation and clustering of fullerenes in the BHJ, thus reducing the donor–acceptor interfacial area, various strategies were proposed involving grafting thermally crosslinkable groups on the fullerene moiety. To our knowledge, acetylenic, styryl, epoxide, oxetane and silyl groups have now been introduced on the fullerene C60core (Fig. 13).

Two examples of fullerene derivatives bearing diacety- lene groups were developed in the search for a thermal crosslinking process. Compound F2D (Fig. 13) was used by Markov et al. in a PPV derivative–poly(F2D) bilayer structure to form an immobilized electron acceptor layer by thermal polymerization,⁴⁴ and another polymerizable derivative composed of four butadiyne subunits was reported by Nierengarten and Setayesh for stabilizing BHJ solar cells.⁴⁵ In the latter case, morphological changes were observed in a poly[2-methoxy-5-(3^′,7^′-dimethyloctyloxy)]–PPV/fullerene ﬁlm occurring during thermal treatment at 100∘C for 2 h with clear indication that crosslinking of this fullerene derivative was able to stabilize the BHJ.

In devices incorporating P3HT donor and C₆₁-butyric acid glyci- dol ester (PCBG) acceptor described by Dreeset al.,⁴⁶the crosslinking of glycidyl ester in PCBG was induced by adding small amounts of the soluble Lewis acid tris(pentaﬂuorophenyl)borane as mild initiator and by heating the sample for up to 1 h at 140∘C. A partial ring-opening polymerization was suggested from infrared spectral studies which showed the disappearance of epoxy ring absorption bands leading to short oligomer chains. While trans- mission electron microscopy (TEM) images of P3HT:PCBM blends showed a strong morphology change with phase separation into large dendrite-like aggregates, P3HT:PCBG ﬁlms remained ‘clear’

indicating no large phase separation because of eﬃcient thermal crosslinking process (Fig. 14). Nevertheless, an unusual degradation of the current–voltage curve was observed during this thermal treatment while the short-circuit current and the open-circuit voltage remained unchanged. An incomplete crosslinking process was suggested to be responsible for this phenomenon.

In order to avoid oxidation of the metal cathode and to suppress the use of PEDOT:PSS layer in conventional devices, Cheng et al.⁴⁷ developed an eﬃcient approach based on inverted solar cell structures in which an inorganic metal oxide (TiO_x or ZnO) layer was inserted between the indium tin oxide (ITO) and the active layer. Considering that surface hydroxyl groups on metal oxide are known to act as eﬃcient electron

(9)

1354

(c) (a)

(b)

Figure 9.(a) Structure of P3HT-azide10. (b) Eﬃciencies of P3HT/PCBM devices containing 0, 10 or 15% of P3HT-azide10 copolymer during thermal annealing at 150∘C with PCBM as the electron acceptor. (c) Current–voltage curves of 0 and 15% P3HT-azide10 devices with initial (3 h) and long-term (40 h) thermal annealing.³¹(Reprinted (adapted) with permission from Kimet al.³¹Copyright 2012 American Chemical Society.)

Figure 10.Stabilization of nanomorphology by solid-state reaction of PCBM with azide-functionalized graft copolymer possessing a graft-chain density of 1% (n=0.01).⁴¹(Reprinted (adapted) with permission from Gholamkhass and Holdcroft.⁴¹Copyright 2010 American Chemical Society.)

traps, the introduction of an additional n-type fullerene interlayer between both ITO and active layers might improve the device performance. Two oxetane-functionalized fullerene derivatives, [6,6]-phenyl-C₆₁-butyric oxetane ester (PCBO) and [6,6]-phenyl-C63-butyric oxetane dendron ester (PCBOD), were synthesized in order to demonstrate that the oxetane func- tionality can anchor onto the TiOx surface through a cationic ring-opening reaction using a catalytic amount of photo-acid generator upon UV irradiation and thermal treatment.⁴⁷Hydroxyl groups on metal oxide can serve as nucleophiles to open the ring of protonated oxetane, thereby forming a monolayer modiﬁer by covalent bonding (Fig. 15).

Using the self-assembly properties of PCBO and PCBOD on TiO_x surface, an inverted BHJ device showed an impressive PCE of 4.1%, whereas the reference device without this monolayer exhibited a PCE of 3.6%. Improvement of exciton dissociation, reduction of charge recombination, decrease of interface contact resistance and passivation of surface electron traps at the interface of TiO_xwere proposed as combined eﬀects for this increase of eﬃ- ciency. The acceptor PCBOD, containing two oxetane groups, was also capable of self-assembling on the TiOxsurface and a corresponding device incorporating this crosslinked PCBOD interlayer showed a highest PCE of 4.5% which represents 26% enhancement over the reference device. The advantages of both self-assembly and crosslinking in a single fullerene-based molecule were then demonstrated.

The same group developed recently a crosslinkable fullerene material, bis(2-(trichlorosilyl)propyl) malonate C60 (TSMC), functionalized with two trichlorosilane groups.⁴⁸Because of an easy hydrolysis of the trichlorosilyl moieties to produce polysiloxane, the TSMC acceptor could be self-assembled and crosslinked on a TiO_x surface by simple spin-coating without the aid of irradiation or post-thermal treatment (Fig. 16). The generation of a self-assembled and crosslinked TSMC layer was shown in particular to lead to an eﬃcient passivation of hydroxyl groups on the TiOx surface and to form a nanostructured surface. This strategy was applied in the fabrication of P3HT:PC₆₁BM- and PDITTDTBT:PC₇₁BM-based inverted devices (Fig. 16). These solar cells showed 22 and 35% improvement of eﬃciency, respectively, over a device without the self-assembled crosslinked TSMC layer.

Cheng and co-workers synthesized a PC61BM-based fullerene derivative that contains a small dendron bearing two thermally crosslinkable styryl groups (PCBSD).⁴⁹ Crosslinking of this PCBSD acceptor was carried out by heating at 160 ∘C for 30 min yielding a robust thin ﬁlm of crosslinked PCBSD (C-PCBSD).

Incorporated as an interlayer in P3HT:PC61BM or P3HT:IC60BA multilayer inverted devices,⁵⁰remarkable PCE values of 4.4 and 6.2%

were obtained, respectively, with an improvement of all photovoltaic parameters, thus corresponding to improvements of 20%

for P3HT:PC₆₁BM and 23% for P3HT:IC₆₀BA model devices. Supe- rior air stability was noted indicating that the C-PCBSD interlayer

(10)

1355

(a) (b)

(c)

Figure 11.(a) Optical microscope images of 4/PCBM (refer to Figure 10) blend (50:50) before (A) and after (B) 1 h treatment at 150∘C and of P3HT/PCBM (50:50) blend before (C) and after (D) thermal treatment (1 h at 150∘C). Scale bar: 100μm. The inset in (D) shows an expanded view (scale bar: 8μm). (b) Current–voltage characteristics of 4/PCBM (40:60 w/w): as-cast ( ), thermally reacted ( ) and dark current ( ). (c) PCE as a function of annealing time for P3HT/PCBM ( ) and 4/PCBM ( ) devices with 40:60 wt% ratios.⁴¹(Reprinted (adapted) with permission from Gholamkhass and Holdcroft.⁴¹Copyright 2010 American Chemical Society.)

S S

S

UV

S S

S

UV

S S

S S S S S S

100 µm 100 µm

Figure 12.Left: schematic of P3HT-A, photochemical anthracene dimerization and reaction between anthracene and PCBM in a [2+2] cycloaddition.

Right: optical microscope images of P3HT-A:PCBM layers after thermal annealing at 150∘C during 2 h (a) without UV treatment and (b) with UV treatment (365 nm) prior to annealing.⁴²(Reproduced by permission of the Royal Society of Chemistry.)

plays a signiﬁcant role in increasing the lifetime of an unencap- sulated device after exposure to air under ambient conditions (Fig. 17).

To extend the eﬀects of the C-PCBSD interlayer on solar cell eﬃciency, a P3HT:IC60BA upper layer was then deposited on a C-PCBSD bottom layer with vertically oriented nanorods with the aim of creating a well-organized nanostructured interface.

A marked increase of PCE reaching 7.3% was achieved and the device exhibited a marked ambient and morphological stability.⁵¹ The results of this A–A approach are currently limited to the functionalization of fullerene C60 with chemical moieties (acetylenic, styryl, epoxide, oxetane and silyl groups) able to polymerize under thermal or photochemical stress or acid initia- tion. This strategy proved to be eﬀective for creating crosslinked networks with suppression of fullerene crystallites in the blends resulting in interesting stabilization of solid-state morphology.

The major drawback lies in the few available polymerizable groups

and the subsequent control of the crosslinking process. To date, no example of a crosslinkable group grafted on the fullerene moiety prone to react with C₆₀itself has been reported in the literature.

Interest should be in the formation of blends of some diads, triads or oligomeric structures with the aim of avoiding the proximity of fullerenes. However, the investigation of this strategy should be limited due to the reactivity of fullerenes as electro-deﬁcient alkenes which involves essentially cycloaddition reactions under thermal or photochemical activation.

To generate a mixed network of donor and acceptor, photo-crosslinking was investigated by Zhu et al.⁵² with the reaction of polythiophene end-capped with epoxide groups and PCBG as an epoxide-functionalized PCBM-based derivative (Fig. 18). This process was initiated by the addition of catalytic amounts of photoinitiator. The thermal stability of P3HT/PCBM (1.67 to 0.85%) and P3HT/PC61BG (1.72 to 0.86%) solar cells was found to be quite similar after 150 min at 120∘C, while the addition

(11)

1356

Figure 13.Crosslinkable fullerene derivatives bearing diethynyl, silyl, epoxide, oxetane or styrene functionalities.

of acid initiator accelerated the polymerization but reduced the eﬃciency of the device (0.9 to 0.0%).

Similarly, a very recent article reports the use of both donor and acceptor materials functionalized with vinyl groups to crosslink on the donor and/or acceptor, so this is an example of the three linking approaches together.⁵³ The donor material was a low-bandgap copolymer based on benzothiophene and thienothiophene (PBDTTT-V) and the acceptor a fullerene derivative (PCBD), as shown in Fig. 19. Diﬀerent blends were considered:

the vinyl-functionalized donor with the classic PCBM, the corresponding electron-donating copolymer without vinyl groups with PCBD instead of PCBM and a blend with both vinyl-functionalized materials. These experiments showed that crosslinking reactions occurred between polymer chains leading to an insoluble ﬁlm.

When PCBM was used, it could washed off the active layer, but this was not the case when PCBD was used, indicating a reaction between fullerene molecules. The most efficient improvement in stability of device performance was obtained when both vinyl-functionalized materials were used, as shown in Fig. 19, where 79% of initial efficiency was retained after 8 h at 150∘C.

Optical microscope images conﬁrmed that fewer crystals are formed compared to PBDTTT-V:PCBM, but fewer than with the non-functionalized materials.

This strategy involving combining the D–D, A–A and A–D approaches⁵³^,⁵⁴has the signiﬁcant advantage of being more general and more versatile even if time-consuming chemistry is required since both acceptor and donor need to be functionalized.

Thus, only a few examples have been reported. Nevertheless, by selecting appropriate functions, this approach may be developed in the future.

To conclude, the three approaches, D–D, D–A and A–A, gave successful results for the stabilization of morphology and the stability of performance over time. However, the different approaches are not equivalent in terms of these parameters. D–D crosslinking usually leads to the poorest stability of the morphology as the diffusion of PCBM is not retained and crystals are observed after long periods of thermal annealing. In contrast, crosslinking between acceptor molecules stabilizes the morphology, but with the trade-off that there is less choice of functions available to per- form reactions onto fullerenes.

(a) (b)

Figure 14.(a) TEM images of P3HT:PCBM (left) and P3HT:PCBG (right) ﬁlms after annealing at 140∘C for 1 h. (b) Current–voltage curves (AM 1.5 simulated, 80 mW cm⁻²) of P3HT:PCBG device post-treatment and after additional thermal treatment (10 min at 140∘C).⁴⁶(Reproduced from Dreeset al.⁴⁶with permission from the Royal Society of Chemistry.)

(12)

1357

O O

O

O SiCl₃

Cl₃Si

TSMC

Hydrolysis Without heat and light

O O

O

O Si

Cross-linked TSMC O

O Si O

O O

O

S R R S

R R

S

NSN S

n

R= _OC

8H₁₇ PDITTDTBT

Figure 16.TMSC and PDITTDTBT polymer structures.

CROSSLINKERS IN POLYMER–FULLERENE BLENDS

The crosslinking approaches previously reviewed all require the synthesis of crosslinkable donor and/or acceptor semiconductors.

Such a material preparation can be costly. These materials may also become unstable upon storage and during OPV processing steps because they are crosslinkable and contain reactive moieties.

Another route to stabilize the BHJ may be the use of crosslinkers, i.e. molecules used as additives, in blends of semiconductors.

This pathway leads to versatility, being applicable to multiple

semiconducting polymers. This aspect is of great importance since the development of OPV-designed high-eﬃciency materials is attracting an increasing amount of interest. In this context, the stabilization strategy has to be applicable to future materials.

Very few examples of crosslinkers have been used in BHJ solar cells so far. Ryu and Jang⁵⁴incorporated an acrylate-based photo-curable crosslinker with a photoinitiator (Fig. 20) into P3HT:PC₇₁BM blends. A slight increase of the device performance was observed (from 3.09 to 3.43% or 3.23% with 1 or 2% of crosslinker, respectively) as a result of the crosslinker, but no stability analysis was reported in that study.

(13)

1358

PCBSD O O

O O

Cross-linking 160°C for 30 min

C-PCBSD O O

O O

n n

Figure 17.Normalized PCE as a function of time for devices A ([ITO/ZnO/C-PCBSD/P3HT:PC₆₁BM (1:1 w/w)/PEDOT:PSS/Ag]) and B ([ITO/ZnO/P3HT:PC₆₁BM (1:1 w/w)/PEDOT:PSS/Ag]) stored for 35 days in air under ambient conditions.⁵⁰(Reprinted (adapted) with permission from Hsiehet al.⁴⁹Copyright 2010 American Chemical Society.)

O

PCBG O O

S C₆H₁₃

n S

m

O O

+

O O

O OH

S C₆H₁₃

S

O

C₆H₁₃ S S

O

O O

S C₆H₁₃

S O I⁺

C₁₂H₂₅ O

OH SbF₆^-

photoinitiator

Figure 18.Schematic of D–A crosslinking process occurring between epoxide-based polythiophene and PCBG derivative.

Heet al.⁵⁵described the use of the crosslinker octane-1,8-diyl bis(1,4-dihydrobenzo[d][1,2]oxathiine-6-carboxylate 3-oxide) (OBOCO; Fig. 21). This small molecule was used as an additive in P3HT:PCBM for crosslinking fullerenes through a heat-triggered Diels–Alder reaction. By adding 5 wt% OBOCO (optimized per- centage) into the OPV ink, devices showed signiﬁcantly improved thermal stability compared with the untreated cells (after 4 days at 150 ∘C). The PCE for the device with 5% OBOCO decreases from 2.77 to 1.72% (by 38%) and that without OBOCO from 2.74 to 0.78% (by 72%). By adding 5% OBOCO into the active layer,

heat-promoted fullerene aggregation was markedly inhibited, leading to signiﬁcantly enhanced thermal stability of the optimal ﬁlm morphology.

In terms of versatile crosslinkers for organic semiconductors, a sterically substituted bis(ﬂuorophenyl azide) (s-FPA) has been developed⁵⁶^,⁵⁷which demonstrated remarkable crosslinking capa- bilities. This compound added in minor amounts (a few weight percent) was able to crosslink polymer chains on their alkyl side chains using a deep-UV trigger (Fig. 22). Researchers have been developing such bis-azide crosslinkers for many years⁵⁸^,⁵⁹because

(14)

1359

O O S

S O

O S S

S S O

O S S

PBDTTT-V

x 1-x

(x = 0,0.16)

O O

PCBD

Figure 19.Left: structures of PBDTTT-V and PCBD. Right: time dependence of eﬃciencies of devices based on PBDTTT:PCBM (1:1.5 w/w), PBDTTT-V:PCBM (1:1.5 w/w), PBDTTT-V:PCBM:PCBD (4:5:1 w/w/w) and PBDTTT-V:PCBD (1:1.5 w/w), under thermal annealing at 150∘C. (Reproduced from Chenet al.⁵³with permission from the Royal Society of Chemistry.)

(a) (b)

S C₆H₁₃

n

P3HT

OCH₃ O

PC₇₁BM

O O O O

O O

O O O

P O

Photoinitiator Photo-curablecross-linker

Figure 20.(a) Structure of photovoltaic cell; (b) chemical structures of P3HT, PC₇₁BM, photo-curable crosslinker and photoinitiator used to fabricate the device.⁵⁴(Reprinted fromSolar Energy Materials and Solar Cells, 94, M-S Ryu, J Jang, Improvement of conversion eﬃciency of bulk heterojunction organic solar cells using photo-curable crosslinker, 1384–1388.)

the UV dose required to trigger this chemistry is low and does not aﬀect electronic and photonic properties of semiconducting polymers.

Pnget al.⁵⁶applied this crosslinking strategy with success to various types of organic electronic devices in organic light-emitting diodes with crosslinking of the emitting layer without aﬀecting electroluminescence properties. The interlayers were eﬃciently crosslinked exhibiting easy multilayer solution processing to avoid dissolution of bottom layers by the solvent used to coat top layers.

Transistors were also shown to be unaffected by this crosslinking procedure. In terms of solar cells, however, s-FPA was not used in polymer–fullerene blends. Solar cells were prepared by backfilling with PCBM a pre-patterned crosslinked polymer layer based on OC1C10-PPV. Patterning was achieved using non-crosslinked phase-directing agents, further developed from the polymer template before backfilling with fullerene derivatives.

Recently, and almost simultaneously, the same team of Ho and co-workers as well as that of Meredith and co-workers used s-FPA as a crosslinker in P3HT sub-layers backfilled with PCBM.⁶⁰^,⁶¹ In both studies, highly efficient OPV cells were produced. Backfill- ing with PCBM such a nanotemplated polymer network may be suitable for large-scale manufacturing. However, as evidenced in

both studies, the resultant BHJ remains sensitive to temperature.

The fullerene molecules are here not locked in the nanostructured polymer template. As a consequence, such a BHJ may not be more stable over time and temperature under operating conditions.

The crosslinking approach has been used to prepare nanostructured donor–acceptor interfaces in photovoltaic cells. Indeed, poly(fluorene-alt-bithiophene) functionalized with oxetane groups blended with polystyrene, in the presence of a photo-acid generator, could be crosslinked via UV light and thermal treatment after deposition.³⁴Surface patterning by removal of polystyrene left a columnar nanostructure (250 nm in diameter) that could be further filled with PCBM. However, devices prepared in this way presented lower PCEs than that of usual BHJ. Finally, in situ n-doping and crosslinking of semiconducting polymers as efficient insoluble electron-transporting materials for inverted polymer solar cells was achieved using a crosslinker similar to s-FPA, namely bis(perfluorophenyl azide).⁶² We recently developed a universal methodology based on a novel bis-azide-based crosslinker able to stabilize any kind of polymer–fullerene blend using thermal crosslinking of fullerene moleculesin situafter com- pletion of the optimized BHJ morphology. This technique enabled

(15)

1360

S O O

OBOCO

OCH₃ O

Cross-linking

O O O

∆ -SO₂

O O

O

O O

S O O

O O

OCH₃ O

PC₆₁BM

O O

O

OCH₃ O

(a) (b)

(c) (d)

Figure 21.Left: chemical structure of OBOCO and crosslinking scheme. Right: evolution of various solar cell parameters with continuous heating at 150

∘C for OBOCO content of 0 and 5%.⁵⁵(Reproduced from Heet al.⁵⁵with permission from the Royal Society of Chemistry.)

Figure 22.Photo-crosslinking of polymer chains with s-FPA.

one to thermally stabilize solar cell eﬃciency of devices based on P3HT:PC₆₁BM as well as of devices based on highly eﬃcient state-of-the-art low-bandgap polymers blended with PC₇₁BM.

CONCLUSIONS AND PERSPECTIVES

Particular attention has been devoted in recent years to the development of new donor and acceptor semiconducting materials, the establishment of relationships between molecular structure and thin-ﬁlm morphology, photophysical and photovoltaic properties, and also the elaboration of new device concepts such as the BHJ photoactive layer to enhance the eﬃciency of organic solar cells.

Among the various organic materials designed and synthesized by chemists, fullerene derivatives are still the most popular and eﬃ- cient acceptor materials used in these devices. But their propensity to phase-segregate and to form crystallites in the active layer is one of the major problems responsible for nanometric morphology instability and reduction in eﬃciency of devices. This review was devoted to the strategy for crosslinking materials as a powerful

method to overcome this drawback. The diﬀerent approaches for crosslinking polymer and/or fullerene materials were reviewed, showing that each of them presents advantages and drawbacks.

In the D–D approach the degree of reticulation can be controlled to allow the structuring of the polymer. Various functions are possible to promote this reaction in the solid state using thermal or UV treatment. The latter is the most interesting, as the crosslinking step can be disconnected from the thermal annealing necessary to optimize the morphology. However, in most cases, PCBM crystallization is not completely eliminated. In the D–A approach, the stabilization of the morphology is improved, and the link to the acceptor can be controlled via the amount of introduced functions. However, the method is limited by the choice of functions available to enable reaction of donor with fullerene. The same remark can be made about the A–A approach.

Moreover, control of addition to fullerenes is important to retain electron-accepting properties. The A–A approach appears to be the most eﬃcient for inhibiting PCBM crystallization upon thermal stress.

(16)

1361

ACKNOWLEDGEMENTS

This review was written as part of a French research project called CEPHORCAS funded by the Agence Nationale de la Recherche (ANR-HABISOL-2010-003). CEPHORCAS is focused on the development of crosslinked materials of long-lived OPV applications.

Each co-author represents each university involved in CEPHOR- CAS. The authors are also grateful to SOLVAY, in particular to Bertrand Pavageau for fruitful discussions.

REFERENCES

1 He Z, Zhong C, Su S, Xu M, Wu H and Cao Y,Nature Photon6:591–595 (2012).

2 Chambon S, Derue L, Lahaye M, Pavageau B, Hirsch L and Wantz G, Materials5:2521–2536 (2012).

3 Rabek JF,Photodegradation of Polymers. Springer-Verlag, Berlin (1996).

4 Gardette J-L, inHandbook of Polymer Degradation, 2nd edition. Marcel Dekker, New York, pp. 671–698 (2000).

5 Rivaton A, Mailhot B, Derderian G, Bussiere P-O and Gardette J-L, Macromolecules3:5815–5824 (2003).

6 Chambon S, Rivaton A, Gardette J-L and Firon M,Sol Energy Mater Sol Cells92:785–792 (2008).

7 Manceau M, Chambon S, Rivaton A, Gardette J-L, Guillerez S and Lemaˆıtre N,Sol Energy Mater Sol Cells95:1315–1324 (2011).

8 Ostwald W,Lehrbuch der Allgemeinen Chemie, vol. 2, part 1. Leipzig, Germany (1896).

9 Dang MT, Hirsch L and Wantz G,Adv Mater23:3597–3602 (2011).

10 Bertho S, Haeldermans I, Swinnen A, Moons W, Martens T, Lutsen L et al,Sol Energy Mater Sol Cells91:385–389 (2007).

11 Swinnen A, Haeldermans I, vande Ven M, D’Haen J, Vanhoyland G, Aresu Set al.,Adv Funct Mater16:760–765 (2006).

12 Treat ND, Malik JAN, Reid O, Yu L, Shuttle CG, Rumbles Get al,Nature Mater12:1–6 (2013).

13 Wang T, Pearson AJ, Dunbar ADF, Staniec PA, Watters DC, Yi Het al,Adv Funct Mater22:1399–1408 (2012).

14 Sivula K, Ball ZT, Watanabe N and Fréchet JMJ,Adv Mater18:206–210 (2006).

15 Renaud C, Mougnier SJ, Pavlopoulou E, Brochon C, Fleury G, Deribew Det al,Adv Mater24:2196–2201 (2012).

16 Mulherin RC, Jung S, Huettner S, Johnson K, Kohn P, Sommer Met al, Nano Lett11:4846–4851 (2011).

17 Zhang Y, Yip H, Acton O, Hau SK, Huang F and Jen AKY,Chem Mater 21:2598–2600 (2009).

18 Bertho S, Janssen G, Cleij TJ, Conings B, Moons W, Gadisa Aet al,Sol Energy Mater Sol Cells92:753–760 (2008).

19 Gevorgyan SA and Krebs FC,Chem Mater20:4386–4390 (2008).

20 Lin Y, Lim JA, Wie Q, Mannsfeld SCB, Briseno AL and Watkins JJ,Chem Mater24:622–632 (2012).

21 Bertho S, Campo B, Piersimoni F, Spoltore D, D’Haen J, Lutsen Let al, Sol Energy Mater Sol Cells110:69–76 (2013).

22 Zuniga CA, Barlow S and Marder SR,Chem Mater23:658–681 (2011).

23 Dautel OJ, Wantz G, Almairac R, Flot D, Hirsch L, Lère-Porte J-Pet al,J Am Chem Soc128:4892–4901 (2006).

24 Sallenave X, Dautel OJ, Wantz G, Valvin P, Lère-Porte J-P and Moreau JJE,Adv Funct Mater19:404–410 (2009).

25 Miyanishi S, Tajima K and Hashimoto K,Macromolecules42:1610–1618 (2009).

26 Kim BJ, Miyamoto Y, Ma B and Fréchet JMJ, Adv Funct Mater 19:2273–2281 (2009).

27 Griﬃni G, Douglas JD, Piliego C, Holcombe TW, Turri S, Fréchet JMJ et al,Adv Mater23:1660–1664 (2011).

28 Waters H, Kettle J, Chang S-W, Su C-J, Wu W-R, Jeng U-Set al,J Mater Chem A1:7370–7378 (2013).

29 Murray KA, Holmes AB, Moratti SC and Friend RH, Synth Met 76:161–163 (1996).

30 Murray KA, Holmes AB, Moratti SC and Rumbles G,J Mater Chem 9:2109–2116 (1999).

31 Kim HJ, Han R, Cho C-H, Kang H, Cho H-H, Lee MYet al,Chem Mater 24:215–221 (2012).

32 Nam CY, Qin Y, Park YS, Hlaing H, Lu X, Ocko BMet al,Macromolecules 45:2338–2347 (2012).

33 Charas A, Ferreira Q, Farinhas J, Matos M, Alcácer L and Morgado J, Macromolecules42:7903–7912 (2009).

34 Farinhas J, Ferreira Q, Di Paolo RE, Alcácer L, Morgado J and Charas A, J Mater Chem21:12511–12519 (2011).

35 Brotas G, Farinhas J, Ferreira Q, Morgado J and Charas A,Synth Met 162:2052–2058 (2012).

36 Carlé JE, Andreasen B, Tromholt T, Madsen MV, Norman K, Jørgensen Met al,J Mater Chem22:24417–24423 (2012).

37 Chen X, Chen L and Chen Y,J Polym Sci A51:4156–4166 (2013).

38 Li W, Chen L, Chen Y, Li F and Yao K,Org Electron13:104–113 (2012).

39 Ouhib F, Tomassetti M, Manca J, Piersimoni F, Spoltore D, Bertho Set al, Macromolecules46:785–795 (2013).

40 Prato M, Li QC, Wudl F and Lucchini V,J Am Chem Soc115:1148–1150 (1993).

41 Gholamkhass B and Holdcroft S,Chem Mater22:5371–5376 (2010).

42 Khiev S, Derue L, Ayenew G, Medlej H, Brown R, Rubatat Let al,Polym Chem4:4145–4150 (2013).

43 Li C-Z, Yip H-L and Jen AK-Y,J Mater Chem22:4161–4177 (2012).

44 Markov DE, Amsterdam E, Blom PWM, Sieval AB and Hummelen JC,J Phys Chem A109:5266–5274 (2005).

45 Nierengarten J-F and Setayesh S,New J Chem30:313–316 (2006).

46 Drees M, Hoppe H, Winder C, Neugebauer H, Sariciftci NS, Schwinger Wet al,J Mater Chem15:5158–5163 (2005).

47 Cheng Y-J, Cao F-Y, Lin W-C, Chen C-H and Hsieh C-H,Chem Mater 23:1512–1518 (2011).

48 Liang W-W, Chang C-Y, Lai Y-Y, Cheng S-W, Chang H-H, Lai Y-Yet al, Macromolecules46:4781–4789 (2013).

49 Hsieh C-H, Cheng Y-J, Li P-J, Chen C-H, Dubosc M, Liang R-Met al, J Am Chem Soc132:4887–4893 (2010).

50 Cheng Y-J, Hsieh C-H, He Y, Hsu C-S and Li Y, J Am Chem Soc 132:17381–17383 (2010).

51 Chang C-Y, Wu C-E, Chen S-Y, Cui C, Cheng Y-J, Hsu C-Set al,Angew Chem Int Ed50:9386–9390 (2011).

52 Zhu Z, Hadjikyriacou S, Waller D and Gaudiana R,J Macromol Sci A 41:1467–1487 (2004).

53 Chen L, Li X and Chen Y,Polym Chem4:5637–5644 (2013).

54 Ryu MS and Jang J,Sol Energy Mater Sol Cells94:1384–1388 (2010).

55 He D, Du X, Zhang W, Xiao Z and Ding L,J Mater Chem A1:4589–4594 (2013).

56 Png R-Q, Chia P-J, Tang J-C, Liu B, Sivaramakrishnan S, Zhou Met al, Nature Mater9:152–158 (2010).

57 Tan Z-K, Johnson K, Vaynzof Y, Bakulin AA, Chua L-L, Ho PKHet al,Adv Mater25:4131–4138 (2013)

58 Khong S-H, Sivaramakrishnan S, Png RQ, Wong L-Y, Chia P-J, Chua L-L et al,Adv Funct Mater17:2490–2499 (2007).

59 Png RQ, Chia PJ, Sivaramakrishnan S, Wong L-Y, Zhou M, Chua L-Let al, Appl Phys Lett91:013511 (2007).

60 Liu B, Png RQ, Zhao L-H, Chua L-L, Friend RH and Ho PK-H,Nature Commun3:1321 (2012).

61 Tao C, Aljada M, Shaw PE, Lee KH, Cavaye H, Balfour MNet al,Adv Energy Mater3:105–112 (2013).

62 Cho N, Yip H-L, Davies JA, Kazarinoﬀ PD, Zeigler DF, Durban MMet al, Adv Energy Mater1:1148–1153 (2011).