Block copolymer strategies for solar cell technology

Abstract

A simple overview of the methods used and the expected benefits of block copolymers in organic photovoltaic devices is given in this review. The description of the photovoltaic process makes it clear how the detailed self-assembly properties of block copolymers can be exploited. Organic photovoltaic technology, an inexpensive, clean and renewable energy source, is an extremely promising option for replacing fossil fuels. It is expected to deliver printable devices processed on flexible substrates using high-volume techniques. Such devices, however, currently lack the long-term stability and efficiency to allow organic photovoltaics to surpass current technologies. Block copolymers are envisaged to help overcome these obstacles because of their long term structural stability and their solid-state morphology being of the appropriate dimensions to efficiently perform charge collection and transfer to electrodes. © 2011 Wiley Periodicals, Inc. J Polym Sci Part B: Polym Phys, 2011

INTRODUCTION

The drive to remove mankind's dependence on fossil fuels, while maintaining stable energy supplies, is of vital importance in the coming decades due to the degradative effects of CO₂ production on the Earth's atmosphere.1 Solar energy is seen as a major contributor to solving these problems; harnessing even a small percentage of the incident solar flux would more than meet total global energy demands.2 Interest in organic photovoltaics (OPVs), especially those based on polymers, has increased exponentially over the last decade because of the use of inexpensive raw materials, the associated ease of processing lightweight, flexible devices, and their inherent installation efficacy.3 The lifetime and power efficiency of inorganic solar cells, on the other hand, far exceeds those of the OPVs. Typically, inorganic devices operate at around 14–19% efficiency,4 whereas the current state-of-the-art organic counterparts have reached a maximum of 8.3% on a laboratory scale.5, 6 Even given the exceptional rapidity of research developments in this field,7, 8 it is still expected that considerable gains in both efficiencies and OPV lifetimes must be made to permit full commercial exploitation and overcome the initial investment costs of high-throughput machinery9 and the economic and political inertia of prior investment into competing technologies such as nuclear energy.9, 10 It is generally thought that once OPVs have reached efficiencies of the order of 10%, coupled with lifetimes of around 10 years, there will be an opening-up of new large-scale markets that can be quickly exploited.11, 12

It is relatively easy to see that the processing steps required to manufacture inorganic solar cells are more energy intensive than those for OPVs, as silicon requires extensive purification and treatment, either through sputtering or crystal growing.13 Moreover, the silicon must be housed within heavy-duty paneling and, therefore, the inorganic cells are somewhat more troublesome to install over large areas. Thus, it is intuitive that the future of solar cell technology relies on significantly reducing the cost of inorganic solar cells and/or increasing the lifetime, processability, and efficiency of OPVs.14 Currently, the latter approach appears to be the most feasible due to the wide variety of synthetic,15 architectural,16 and processing17, 18 options available to optimize these systems.

The principal difference between inorganic and organic photovoltaic devices is the exciton binding energy, where an exciton is a Coulombically bound electron-hole pair.19, 20 Inorganic materials have low exciton binding energies such that photoexcitation spontaneously creates a separate free electron and a free hole (a hole being a positive charge located within the material's highest occupied molecular orbital, HOMO), which can both travel directly to their respective electrodes. OPVs, on the other hand, have higher exciton binding energies and, therefore, excitons must reach a material interface with a lowest unoccupied molecular orbital (LUMO) offset to produce separated electrons and holes.21

The photovoltaic mechanism in organic devices has been discussed at length in the literature,19, 22–29 therefore, here, we only briefly outline the process for readers new to the field. Figure 1 shows the individual steps involved in the dominant process that converts light into an electrical current: (i) the incoming photon excites an electron from the HOMO to the LUMO of the donor material to (ii) create an exciton, (iii) which traverses the donor material to a donor–acceptor interface where (iv) the excited electron separates from its bound hole onto the LUMO of the acceptor. Subsequently, (v) the free electron and hole travel through the acceptor and donor materials (vi) to reach the cathode and anode, respectively. It is the continuous percolation of electrons and holes across the device that generates the electric current.

Figure 1. The photovoltaic mechanism involved in organic solar cells: (i) a photon is absorbed by the donor material, (ii) to excite an electron from the HOMO to the LUMO, creating an exciton. (iii) The exciton travels through the donor material to the donor/acceptor (D/A) interface, where (iv) the electron passes to the LUMO of the acceptor material, breaking up the exciton into a free electron and hole. (v) The electron travels through the acceptor material, whereas the hole is passed through the donor material. (vi) Finally, the disparate charges arrive at their corresponding electrodes to create a potential gradient across the device.

The process depicted in Figure 1 describes how incident photons generate electricity in OPV devices, neglecting unsuccessful pathways. Major ways in which OPVs fail include photon loss (poor absorption of the solar spectrum), exciton loss (recombination before charge separation), and charge carrier loss (recombination before arrival at the electrode) as well as the formation of nonradiative excitons (energy lost as phonons).11 Current research focuses on minimizing these losses to maximize power output of the devices. There are several parameters which describe the performance of solar cells and the most useful parameters used throughout the literature for comparing devices are explained below.

The ratio of the collected electrons to incident photons at a specific wavelength is described as the external quantum efficiency (EQE).3 This parameter depicts the fraction of photoexcitations that result in useable charge carriers at the electrodes.30 The internal quantum efficiency (IQE) is the ratio of collected charge carriers to absorbed photons of a specific energy.31, 32 IQE values for OPVs are close to 100%, whereas EQE values will always be lower (although often greater than 80%). The short-circuit current density (J_sc) is the current at zero bias (i.e., when there is no potential difference across the device) and is related to the amount of absorbed photons and the charge carrier mobility of the materials within the device. Another intrinsic device parameter used in conjunction with J_sc is the open-circuit voltage (V_oc), which is the voltage produced when the current in the cell equals zero26, 23, 30 and has been shown to be directly related to the difference between the HOMO of the donor and the LUMO of the acceptor.26 A larger difference between the two levels will give rise to a larger maximum theoretical V_oc for the device; however, the larger the band gap of the donor material, the poorer the overlap between the device's absorption and the solar spectrum (which will affect the overall power conversion efficiency, PCE, or η). Finally, the fill factor (FF) is the ratio of the maximum power produced to the product of the J_sc and V_oc. The FF describes the shape of the current–voltage characteristics of the solar cell and the higher the value, the better the device.23, 30 Equation 1 directly relates η, FF, V_oc, J_sc, and the incident power (P_in) and Equation 2 gives the FF as a ratio of the voltage (V_max) and current (J_max), which gives the maximum power output over the V_oc and J_sc. There are a number of mathematical derivations which further link all of these parameters; however, this is beyond the scope of this review and the reader is directed elsewhere.3, 23

• (1)

• (2)

In short, FF is an indication of the efficiency of charge collection at the respective electrodes and so illustrates the connectivity of the pathways between the electrodes. J_sc is related to the amount of absorbed photons and the effectiveness of the layer to generate free charges and collect them at the electrodes, whereas V_oc is dictated by the energy gap between the HOMO of the electron donor and the LUMO of the electron acceptor, with a reduction of about 0.3 V due to the energy penalty associated with free carrier formation. The overall PCE (η) is directly affected by J_sc, V_oc, FF, and the power input, P_in.

The use of block copolymers to improve overall device efficiency of OPVs has become a hot topic in recent years for a variety of reasons. In this review, we highlight these factors after providing a brief overview of the work performed on “classic” organic solar cells, that is using homopolymer blends (rather than covalently connected copolymers). Following on from this, the self-assembling behavior of block copolymers is discussed before focusing on the different strategies taken to synthesize, utilize, and process block copolymers. There are a number of reviews in the literature that cover various aspects of the use of block copolymers in optoelectronics33 and photovoltaics,34 and also the challenges faced when using block copolymers in such devices,35 the various active donor–acceptor systems used,36 conjugated block copolymer behavior,37–39 and nanostructural considerations.40 Here, we have constructed a review for readers new to the field of OPVs in an attempt to stimulate progress toward highly efficient block copolymer-based devices. It is important to note that this review does not concern alternating copolymers based on donor–acceptor units (the so-called push–pull low-band gap polymers) as these are not block copolymers, and the reader is directed elsewhere for such work.15, 41–43

BRIEF OVERVIEW OF BULK-HETEROJUNCTION OPVs

OPVs based on the bulk-heterojunction (BHJ) structure have been well discussed elsewhere, so only points salient to this review will be discussed. Excellent reviews and updates have appeared over the last few years which trace the extremely rapid development of this field, most notably those by Dennler et al.,3 Günes et al.,7 Brabec et al.,11 and Deibel et al.,23 along with that of Clarke and Durrant that gives an overview of the physics of charge generation and transfer.24 It is generally recognized that in addition to the effects of the qualities of solar spectral absorption, the physics of charge transfer, positioning of bands of the respective donor and acceptor groups, architecture and nanomorphology of the optoelectronically active layer all play pivotal roles in determining the overall success of a device. The change in OPV structure, from a bilayer, shown in Figure 2(a), to the BHJ structure drawn in Figure 2(b), facilitated a step change in attainable efficiencies. It is expected that the structure shown in Figure 2(c) will permit a similar step change, once the underlying nontrivial chemistry and technological details have been overcome. Recent explorations of this are discussed in the following sections: Synthesis, Block Copolymers in the Optoelectronically Active Layer, Processing, and Current Best Devices.

Figure 2. Representations of the structures of: (a) bilayer; (b) BHJ; and (c) block copolymer-based devices. All three have transparent front electrodes (often indium tin oxide, ITO, as an anode, top), and reflective back-electrodes (such as aluminium as the cathode, bottom). In the case of (a), no additional smoothing layers were present in initial devices. In (b) and (c), electron and hole blocking layers, such as the transparent PEDOT-compl-PSS and LiF, top and bottom respectively, are used. Typically, the substrates (bottom one not shown for clarity) are based on flexible polymers such as poly(ethylene terephthalate) (PET) or glass.

The photovoltaic effect was first observed for conjugated polymers in the work of Sariciftci et al.22 While this was an enormous revelation, there were obvious limitations to the effectiveness of the bilayer devices detailed in the article submitted in the same year by the same group44 and shown in Figure 2(a), wherein an efficiency of 0.04% was declared. Although the term “heterojunction” was used in this article, it simply concerned the junction between two differing materials. These were layers made by sequential casting of a poly(phenylene vinylene) (PPV) based donor and vacuum deposition of the fullerene (C₆₀) acceptor. As excitonic states can only exist over a distance of between 4 and 20 nm in these materials,45 any states outside of the interfacial zone are lost, thus resulting in the low efficiencies observed. Furthermore, there is an inverse relationship between the thickness of the material (to improve optical absorbance) and the ease of photocurrent transfer across the device (due to internal resistances) that greatly limited any advances in bilayer device efficiencies. A step change in efficiencies occurred on moving to the BHJ design shown in Figure 2(b). This was due to the increase in interfacial contact area between the donor and acceptor domains, as detailed in the patent literature46 filed in 1994, and elaborated in the literature using a PPV-derivative as a composite with phenyl-C₆₁-butyric acid methyl ester (PCBM).47 Independent work using a composite of donor and acceptor polymers to give a BHJ also pioneered this technique.48, 49 The casting, from solution, of a mixture of donor and acceptor materials to give a thin-film composite generally around 100 nm thick is common to these systems. It should be noted that both electronic and optical parameters play an important role in determining the optimum film thickness.31 During casting and subsequent annealing processes, there is a process of crystallization, alignment of polymer chains, and limited self-organization, which changes the structure to favor exciton collection and charge transport.50, 51 This process is extremely sensitive to the solvent used,52, 53 variations in composition,54 annealing,55–59 chemical-structure variations,60–63 purities,64 and processing conditions,65 to name but a few of the related parameters when dealing with what is a fundamentally unstable system.66 The seminal work by Yang et al. demonstrated the necessity of controlled mixing of the acceptor and donor in the composite. Crystalline zones transport charges away from their point of formation, which often occurs in amorphous zones where the donor and acceptor are more closely intermingled.67 Evidently, the aforementioned parameter of the mean pathway length of an excitonic state for any particular material determines to a great extent what the optimum composite blend might look like.21 Most of the processing and structural parameters that have been mentioned above are essentially concerned with optimizing the morphology around this important parameter and that of charge transfer.21, 68–72 Several of the more common organic chemical structures that have been used in BHJ devices by various researchers are shown in Figure 3. These materials are frequently referred to throughout this review.

Figure 3. Some of the common chemical structures recently used in the development of OPVs and mentioned throughout the text.

In terms of the materials used since the original use of PPV-derivatives, it was found that polythiophenes were considerably more stable.73, 74 Poly(3-hexylthiophene) (P3HT) in particular, has become a standard due to its ease of preparation and handling,75, 76 its reasonable correlation with the bands of the common acceptor PCBM and, while limited to higher energies, correlation with the solar spectrum.26 More recent work has tended to exploit the so-called push–pull low-band gap polymers, as briefly mentioned in Introduction Section, which exploit alternating structures based on acceptor and donor units to reduce the band gap to closer proximity with available light. The reader is directed to the excellent reviews by Leclerc and coworkers41 and Cheng et al.15 on these materials.

Work has not stood still on the side of the n-type material. Improvements relative to PCBM,77 which has been exceptionally successful as a processable acceptor since its inception, have been made. However, such progress has been at a slower pace than the aforementioned low band-gap polymers, perhaps due to the inherent tendency of polymers to prefer hole conduction28 and the difficulties of working with weakly soluble fullerenes.78 Phenyl-C₇₁-butyric acid methyl ester (PC₇₁BM) has proved useful79 due to its increased light absorption at higher wavelengths with respect to PC₆₁BM.80 Of particular note are the “bisPCBM” acceptors, which demonstrated that the fullerene sphere can actually be improved as an acceptor, due to band modifications, when substituted more than once.81 Subsequent work on indene-substituted C₆₀ have proved to be of great interest.82 To rectify one of the problems with PCBM (namely its tendency to undergo excessive crystallization), research has effectively demonstrated the use of cross-linking groups.83, 84 Of similar interest are polymers carrying pendent fullerenes,85 although they tend to suffer from aggregative effects of C₆₀ rather than self-assembly,86 or linear main-chain polymers based on C₆₀ as a monomer, which seem to have displayed morphologies more appropriate to charge collection and transfer.87 Further work on other acceptors such as perylene, notably perylene tetracarboxydiimide (PDI),88, 89 the poly(benzimidazobenzophenanthroline) ladder (BBL),90 and units based on 9,9′-bifluorenylidene,91 conversely, have shown that fullerene may not be necessary.

BHJ cells are generally prepared by the sequential casting (or indeed ink-jet printing18) of a water-based dispersion of poly(2,3-dihydrothieno[3,4-b][1,4]dioxine)-compl-poly(vinylbenzenesulfonic acid) (PEDOT-compl-PSS) on a transparent indium tin oxide (ITO) anode, a near equi-mass solution of donor and acceptor materials in a solvent such as dichlorobenzene, and then the whole is topped-off with an evaporated aluminium cathode. The PEDOT-compl-PSS layer smoothes the ITO surface and facilitates hole conduction, resulting in improved efficiency.92 As shown in Figure 2(b), there is often a layer of LiF between the aluminium and the conducting layer. It is often less than 1 nm thick, and is thought to contribute to reducing the workfunction of the electrode. Currently, however, its role is not completely understood.93 In morphological terms, it is noticeable that there is an influence over the vertical profile of the donor/acceptor composite by the substrate, as would be expected given the different surface energies of the various layers in the device.29, 94, 95

Blend systems are currently limited due to their morphological instability. The desired conformation, depicted in Figure 2(c), is kinetically-trapped in place and thermal migration of the domains is known to occur with time. Aggregation of like domains causes disruption to the nanomorphology, creating microscopic (and even macroscopic) domains of acceptor and donor material. This phenomenon reduces the interfacial area between donor and acceptor, increasing the pathway length an exciton must travel before charge separation can occur. One option to stabilize the BHJ morphology is to turn to block copolymers, which undergo a thermodynamic self-assembling process known as microphase separation.

MICROPHASE SEPARATION

Block copolymers segregate into disparate domains of the polymer constituents on the nanometer length scale due to the inherent incompatibility of the polymer segments.96, 97 The material, however, cannot macrophase separate owing to the covalent linkage that binds the polymer blocks together. This self-assembly is referred to as microphase separation, which is perhaps somewhat misleading as the domain sizes are in the order of nanometers, rather than micrometers.

An important thermodynamic parameter that describes the miscibility of two polymer entities is the Flory-Huggins parameter (χ).98, 99 A negative value of χ indicates favorable mixing, whereas positive χ values arise from unfavorable interactions between the different polymer segments.100 More pertinently for block copolymers, the product, χN (where N is the copolymer total degree of polymerization), controls the state of microphase separation as it represents the segregation strength of the two chemically disparate polymer constituents. A weakly segregating polymer pairing which has χN < 10 will generally favor an isotropic, disordered morphology as the thermodynamic driving force to self-assemble is not sufficient to overcome the entropic stretching penalty associated with demixing. Generally, when χN > 10.5, microphase separation occurs and a range of structures can be observed depending on the exact value of χN and the relative volume fractions (f) of the polymer blocks. It is important to note at this point that as f deviates from 0.5, higher χN values are required for microphase separation to occur.

There are two major types of polymer chain to consider when discussing microphase separation: coils and rods. A coil describes a flexible polymer chain that prefers to adopt an amorphous “random walk” conformation. Conversely, a rod describes a polymer chain which creates a rigid, crystalline, unidirectional block, generally due to overlapping π-orbitals. In optoelectronic applications, the presence of rod blocks is crucial as they facilitate charge transport due to their π-conjugated systems. However, the presence of coil or rod blocks has tremendous implications on the self-assembly process; therefore, both chain flexibility and electronic properties must be taken into consideration when designing photovoltaics from block copolymers.

Coil–Coil

Coil-coil block copolymers are the most conventional type of diblock copolymer and their self-assembling behavior has been studied for several years.96, 100–102 Their phase behavior is solely driven by the immiscibility between the disparate polymer segments against the entropic penalty associated with chain stretching. There are a distinct series of nanostructures adopted by coil-coil block copolymers. The precise equilibrium nanostructure that will be adopted by a specific block copolymer will be dictated by χN and f. Figure 4 shows the most common morphologies that are adopted by coil-coil diblock copolymers.

Figure 4. Schematics of the most commonly observed nanostructures adopted by coil-coil block copolymers, where S = spheres, C = cylinders, G = gyroid and L = lamellae. The blue block indicates polymer A (with volume fraction, f_A) and the red block polymer B.

Control over the relative volume fractions of the polymer blocks (f) provides access to a series of nanostructures. When f ∼ 0.15, spheres (S) of the minority polymer component will be formed, surrounded by a continuous matrix of the majority polymer block. Increasing f to within the range of 0.2–0.35 will produce hexagonally packed cylinders (C, often referred to as HEX) of the minority component, separated by a continuum of the majority block. A further increase in f (0.35–0.4) would favor a gyroid (G) morphology, where the minority component persists within the inner domains of the curvature, before a lamellae (L) structure becomes the most thermodynamically stable for symmetrical diblock copolymers (f = 0.5). It is important to note that the volume fractions suggested here are by no means confined to these values. Different polymer pairings will give rise to these morphologies over varying f ranges, depending on their specific χ values. Clearly, a morphology where there are isolated spherical domains would not be suitable for OPV applications, as they do not provide continuous polymer pathways for the electrons or holes to travel to the electrodes. Conversely, all of the other morphologies adopted by coil-coil block copolymers would allow percolation of electrons and holes, provided that the domains were appropriately aligned perpendicular to the substrate and that charge-carrier mobilities were sufficiently high. There are other morphologies which have been observed in block copolymers, such as perforated lamellae and double diamond structures.103 From a design perspective, these structures would be ideal for OPVs as they provide continuous pathways to both electrodes. However, such bicontinuous mesostructures are only observed within a small window (compositional and temperature ranges) of the phase diagram and are not commonly observed in all diblock copolymers, thus they are difficult to target.

Conjugated Rod-Based Block Copolymers

When discussing solar cell applications, the inherent problem with amorphous coil-coil block copolymers is their extremely low charge carrier mobility (as there are no overlapping π-orbitals).104 Consequently, coil-type polymers cannot be used as the photoactive materials themselves, but may serve as templates, scaffolds or compatibilizers for the conducting materials (see Block Copolymers in the Optoelectronically Active Layer section). For photoactive polymers, we must turn our attention to rod-based polymers. The two major options available to block copolymer engineers in this field are (i) rod-coil block copolymers, with pendant conductive groups placed along the flexible backbone of the coil segment or (ii) rod-rod block copolymers, which have high charge-carrier mobility in both blocks, but lack conformational flexibility, as discussed herein.

Rod-containing block copolymers have a more complex phase diagram than classical coil-coil block copolymers as there is an additional driving force to the order-disorder transition (microphase separation) from the crystallization of the rod blocks.105 These driving forces are in competition with one another; the most dominant of which is dictated by the segregation strength (χN), the relative volume fractions of the polymer blocks (f), the propensity to crystallize (linked to the Maier-Saupe constant, ω, which describes the rod-rod alignment where steric repulsion favors liquid crystal formation) and the thermal history of the polymer. Rods have a higher propensity to form planar domains, rather than rounded morphologies. Consequently, more phase space is covered by the lamellae or liquid-crystalline structures, rendering it more difficult to target nonlamellae-type morphologies from rod-containing block copolymers.106, 107

For rod-based block copolymers, it is the ratio ω/χ that plays an important role in determining which microstructure is adopted. This ratio describes whether microphase separation or liquid crystalline behavior will dominate. When ω/χ is greater than unity (i.e., ω > χ), alignment of the rod blocks will dominate and the observed morphologies will be those resembling liquid crystals, whereas values lower than unity are more likely to produce nanostructures akin to coil-coil diblock copolymers. Even when ω/χ < 1, the presence of rod blocks favors the formation of “sheets,” rather than rounded domains thus more of the phase diagram is occupied by the lamellae phase rather than spherical or cylindrical morphologies for rod-coils. When the rods prefer to align (ω/χ > 1), the lamellae regime becomes overwhelmed by a smectic phase. The most commonly observed liquid crystal phases that are adopted by rod-coil block copolymers are illustrated in Figure 5.38 A structure which is not analogous to those adopted by coil-coil block copolymers has been observed in rod-coil block copolymers at high coil contents (f_coil ∼ 0.9). The so-called hockey puck morphology consists of a cylindrical disc of rod blocks with flexible coil chains emanating from the surface, as shown in Figure 5(e).108 Such isolated domains, however, would not be suitable for photovoltaic applications due to the absence of continuous pathways to the electrodes, as discussed above. Detailed discussion of the synthesis, self-assembling behavior and applications of rod-based block copolymers can be found elsewhere.37–39, 105, 109

Figure 5. Illustrations of rod-coil block copolymer self-assembly into (a) nematic, (b) bilayer smectic A, (c) monolayer smectic A, (d) monolayer smectic C and (e) “hockey pucks.” Adapted from de Cuendias et al.38 Polymer International (2010) with permission from Wiley.

When rod-based block copolymers are extensively annealed, long range order is commonly observed in the form of nanowires or fibrils.110–113 The individual wires consist of aligned rod blocks and have been shown to be particularly valuable for charge mobility. It should be stressed, however, that all of the aforementioned nanostructures are dependent on the processing history of the polymer. Parameters such as solvent used, annealing conditions, shear force and film geometry all affect the final morphology, particularly for rod-based block copolymers.

Finally, thin film self-assembly is particularly pertinent in solar cell technology. Surface effects from the substrate can influence the nanoscale morphology in a number of ways. For example, surfaces that are preferential to one of the polymer blocks tend to drive the formation of polymer islands, whereas nonselective surfaces typically lead to nanostructured films.114, 115 However, fine-tuning of the surface selectivity can also drive the formation of nanodomains which are parallel or perpendicular to the substrate, depending on the strength of the substrate field.116 Additionally, many groups have reported on the controlling of the final morphology by the careful addition of homopolymer to the diblock copolymer in the melt, where the homopolymer is chemically identical to one of the segments of the block copolymer.117 In particular, this approach has been shown to drive the formation of perpendicularly arrayed nanopillars.118 Clearly, this is extremely useful for solar cell applications, where continuous (i.e., perpendicular) pathways between the electrodes are essential. For a more comprehensive overview of block copolymer self-assembly in thin films, the reader is directed elsewhere.119

SYNTHESIS

There have been a number of different routes used to synthesize the block copolymers featured in this review. Here, we very briefly compare and contrast some of the most common approaches and suggest strategies for the future large scale manufacture of such materials. The reader is directed elsewhere for more in depth reviews on the synthesis of conjugated polymers15 and rod-coil conjugated polymers37.

The synthesis of block copolymers can be achieved by one (or a combination) of two main approaches; “step growth-like” and “chain growth-like” syntheses (see Fig. 6). A step growth-like approach involves the synthesis of the individual segments of the block copolymer first, before a coupling step (or steps) to covalently bind the constituent blocks together. The advantage of this approach is that the individual chemistries to produce the segments can be fully optimized to create well-defined macromolecules with predetermined molecular weight. However, the use of coupling steps can prove troublesome when they are not quantitative, which is often the case when working with macromolecules with only one distinct functional group on each chain that is available for reaction. Consequently, a step growth-like approach requires highly efficient coupling chemistry, such as click chemistry,120 otherwise the eventual yield of the block copolymers will be low.121–123 Clearly, low yield of the final active block copolymer puts serious doubt on their industrial viability.

Figure 6. A comparison of the step and chain growth-like approaches for the syntheses of rod-coil block copolymers.

The most convenient, and indeed most common, approach is the chain growth-like method. An individual block is polymerized and subsequently functionalized to incorporate a group capable of initiating a controlled polymerization. The afforded polymer is aptly referred to as a macroinitiator. Currently, the most common macroinitiators used to produce block copolymers for use in solar cells come from the poly(alkylthiophene) (P3AT) family. The most effective method used to produce P3AT macro-initiators is Grignard metathesis (GRIM) followed by a relatively straightforward end-group conversion.124 GRIM was pioneered by Yokoyama et al. and Sheena et al. and is now the most popular method for producing appropriate regioregular P3ATs with low molar mass dispersities.75, 125–127 P3AT macro-initiators (synthesized via GRIM or other methods) have been end-functionalized to create block copolymers via atom transfer radical polymerization (ATRP),128–132 reversible addition-fragmentation chain transfer (RAFT),133–135 nitroxide mediated radical polymerization (NMRP),133, 136–143 anionic polymerization144, 145 and ring-opening polymerization (ROP).146

These polymerization techniques are capable of yielding a wide range of well-defined polymers and have advantages and disadvantages associated with them, as extensively covered across the literature.147–151 However, the most pertinent drawback concerning subsequent use for solar cell applications are those polymerizations and “click” coupling reactions which rely on metal catalysts, such as copper, with ATRP being the prime example. Such metallic impurities must be completely removed from the final block copolymer otherwise they interfere with the optoelectronic properties of the device.152

The most common approach for producing rod–rod donor–acceptor block copolymers with a completely conjugated rigid backbone are polycondensation reactions. These are numerous in type and will not be discussed in detail here so the reader is directed toward selected examples in the literature and references therein.153–156 The simplest approach for synthesizing donor–acceptor block copolymers, however, currently relies on the polymerization of a flexible, nonconjugated backbone with pendent conjugated repeat units, which display π-π stacking with their nearest neighbors. This allows controlled radical polymerization to be used for both blocks as exemplified with NMRP by Thelakkat's group in 2007 to yield the poly(4-vinyltriphenylamine)-block-poly(perylene bisimide acrylate) (PvTPA-b-PPerAcr) shown in Figure 7(a) (where Figure 7 shows a wide range of block copolymers discussed in this review).152 The minor drawback with this approach is the synthetic complexity involved in preparing the conjugated monomer units, particularly the acceptor moiety which, in this example, was based on perylene bisimide. In this respect, the same group has used NMRP for both blocks to synthesize conducting polymer segments attached to conventional coil blocks [4-vinylpyridine in Fig. 7(b)157 and styrene in Fig. 7(c)158] toward use in solar cells. Sequential GRIM has mainly been used to produce donor-donor conjugated block copolymers based on P3ATs,104, 159, 160 and has recently been extended to polythiophene blocks carrying pyridine groups to increase interactions with PCBM.161 However, Izuhara and Swager have used the technique to produce donor–acceptor type block copolymers by using a final Yamamoto-type cyclization to produce the acceptor block (as shown in Fig. 8).162

Figure 7. Example block copolymer structures as detailed in the text: (a) PvTPA-b-PPerAcr;152, 177, 178 (b) poly[N,N'-bis(4-methoxyphenyl)-N-phenyl-N'-4-vinylphenyl-(1,1′-biphenyl)-4,4′-diamine]-block-poly(4-vinylpyridine) (PVDMTDP-b-P4VP);157 (c) polystyrene-block-poly(perylene bisimide acrylate) (PS-b-PPerAcr);158 (d) poly{(1,4-fullerene)-alt-[1,4-dimethylene-2,5-bis(cyclohexyl-methyl ether) phenylene]}-block-poly(3-hexylthiophene) (PFDP-b-P3HT);123 (e) structure based on norborene units carrying pendent fullerene and P3HT [denoted in the original work poly(1)-block-poly(2) and here PP3HTNor-b-PC₆₀Nor to aid differentiation];179 (f) poly(3-hexylthiophene)-block-poly(bisimide perylene acrylate) (P3HT-b-PPerAcr);180 (g) block copolymer based on a P3HT block and an acrylate/styrene copolymer carrying PCBM units [P3HT-b-P(S_xA_y)-C₆₀];135 (h) poly(3-hexylthiophene)-block-poly(perylene diimide acrylate), the latter block also being known as poly(perylene bisimide acrylate) (P3HT-b-PPDA);143 (i) poly(3-hexylthiophene)-block-poly(perylene bisimide acrylate)s of varying molecular weights (P3HT-b-PPerAcrs);181, 182 (j) P3HT-based block copolymers carrying pendent PCBM moieties (P3HT-b-(PCBM-P3HT)].124

Figure 8. Poly(pyridinium phenylene)-based acceptor blocks in the main-chain. Reprinted with permission from Izuhara and Swager.162 Copyright 2011 American Chemical Society.

While device efficiencies and stabilities remain low, material costs should be as low as possible.163 This has two effects. First, the costs of production (through materials, solvents etc.) will need to be as low as possible to make block copolymers commercially viable. Second, the materials will have to demonstrate added value, e.g. greater efficiency, strength, stability etc. in the final device that is commensurate with their increased costs so that the market can be persuaded to pay more. While block copolymers might ultimately reduce costs by simplifying production lines (as there is only one component), the costs of their preparation may be more expensive than typical blends, and this in turn will mean that block copolymers will have to demonstrate better properties than simple blends to be commercially viable. This should not be considered an obstacle to their development though. Block copolymers are already widely used in commercially viable processes because they possess the properties that the market-place considers necessary. For example, Styrolux® (a styrene-butadiene block copolymer) is produced at the rate of 110000 tons per year.164

BLOCK COPOLYMERS IN THE OPTOELECTRONICALLY ACTIVE LAYER

The ways in which block copolymers are used in OPV active layers can be divided into three classes: (i) as the main component in the layer, either as received or with additives; (ii) as a minor component acting as a compatibilizer to improve and/or stabilize the blend structure; and (iii) as a template for the optoelectronically active structures. The three are considered in the following subsections.

As Single and Multicomponent Layers

This field has recently been reviewed from varying perspectives by Darling,34 Segalman35 and de Cuendias38 so, therefore, only a brief sketch of the various strategies taken so far will be given. As detailed above, the main reasons that performance in OPVs are considerably lower than those of their silicon cousins are based on losses of charge-carriers (through recombination) and excitons, and mismatches between the bands of the chromophore, the solar spectrum and the components in the device.165 Band-mismatches are currently being worked on by varying repeat-units in the polymers being used, but this remains complementary to work being performed on excitonic and charge former losses as they can be strongly influenced by geometrical and architectural changes in the device structure.

While we are more used to seeing nature be cruel with mismatches in parameters, she has perfectly matched the typical dimensions of block copolymers and excitonic states. That the size of the domains formed by block copolymers, described in Microphase Separation section, are typically of the order of the mean pathway length of an excitonic state (typically between 5 to 20 nm during its lifetime depending on the material),45 is an incredible coincidence. If it can be exploited by making materials that can both capture excitons (by providing an interface every 10 nm or so) and facilitate charge percolation through channels while minimizing recombination, then the route to devices with massively improved efficiencies might be found.

One of the first works to target this ideal device was performed by Stalmach et al.136 While they did not demonstrate the material in an OPV, they showed the fundamentally important fast electron transfer from a donor block to an acceptor block. A PPV-derivative was prepared via the Siegrist polycondensation reaction to obtain a formyl group at a chain-end, which was then reacted with a small functional unit to prepare a (2,2,6,6-tetramethyl-piperidin-1-yl)oxyl (TEMPO)-based macro-initiator for the subsequent copolymerization of styrene and 4-chloromethylstyrene by NMRP. Modification of the thus-formed polystyrene (PS) segment by addition of fullerene to the chloromethyl-groups yielded the copolymer shown in Figure 9 with a high concentration of fullerene.

Figure 9. Structure of PPV-block-P(S-stat-C₆₀MS) from Polymer, 42, de Boer et al. Supramolecular self-assembly and optoelectronic properties of semiconducting block copolymers, pp 9097–9109.138 Copyright (2001), with permission from Elsevier.

Subsequent work138 indicated that the low solubility of the PPV derivative, albeit of what might generally be considered oligomeric size, hindered the polymerization of the styrenic groups. Nevertheless, the authors obtained a high number of C₆₀ molecules on their polystyrene block, with around one C₆₀ for every two styrene units. More importantly, the authors demonstrated both a clear formation of domains of each block, as shown in Figure 10, and most excitingly, that the block copolymer gave a greater efficiency as an active layer in a photovoltaic device than the simple blend of the founding blocks.

Figure 10. AFM data (tapping mode, 400 x 400 nm) of poly(phenylene vinylene)-block-polystyrene [(a) and (b)] and PPV-block- P(S-stat-C₆₀MS) [(c)and (d)]. Images (a) and (c) are reprinted from Polymer, 42, de Boer at al., Supramolecular self-assembly and optoelectronic properties of semiconducting block copolymers, pp 9097–9109.138 Copyright (2001), with permission from Elsevier. Images (b) and (d) are three-dimensional illustrations adapted from images (a) and (c) using POVRAY software to emphasize morphological, rather than topographical, variations within the film.

Later work with a comparable block copolymer, this time with a higher molecular weight PPV-derivative, suggested structures close to the ideal vertically organized lamellae perpendicular to the substrate, as in Figure 2(c).140 However, a consistent problem in this work was the way in which the C₆₀ tended to strongly influence the structure of the device. This was later addressed by preparing a block copolymer from P3HT and poly(4-vinylpyridine) (P4VP), with the latter block being used to selectively aggregate PCBM, as shown in Figure 11(a), through noncovalent bonding within its domains.166 It was also found that the way in which the block copolymer was processed (discussed in Processing section) and used in the final device structure was extremely important. When a “normal” device structure was used, that is the P3HT-block-P4VP-blend-PCBM cast onto a PEDOT-compl-PSS substrate (Fig. 11), there was a high concentration of P4VP blocks at the interface, resulting in a counter-diode effect in the I-V curve, and a reduced efficiency of 0.03%. Conversely, using an “inverted” device structure,71 wherein the polymer was cast onto the cathode, the effects of preferential wetting was negated and a considerably higher efficiency of 1.2% was found. The idea behind this work can be compared to the aforementioned work of Maria et al.157 where a similar methodology was explored using a p-type block carrying tetraphenylbenzidine pendent groups and a P4VP block [see Fig. 7(b)]. This latter polymer was used to selectively sequester CdSe:Te nanocrystals during the casting phase. Interestingly, the size of the P4VP domain did not change with the incorporation of the CdSe:Te.

Figure 11. (a) The structure of P3HT-block-P4VP-compl-PCBM and its use in (b) “normal” and (c) “inverted” OPV architectures. Reprinted from Sary et al.166 Advanced Materials (2010) with permission from Wiley.

The parameters of molecular design and 3-dimensional effects on exciton-collection and charge percolation to the electrodes have been extensively discussed by Sun167 in terms of the still low efficiencies for such devices with respect to the BHJ-based devices being fabricated. While some of these effects could be blamed on mismatches in bands, the essential differences arose from the researchers being obliged to concentrate on “simply” demonstrating that block copolymers could work, rather than choosing materials that might optimize their operation. In effect, and this is something that has remained a handicap for block copolymers, the materials that arose were more to do with the available synthetic pathways than the eventual properties of the thus-obtained materials. For example, the polystyrene used in the works by Hadziioannou and coworkers detailed above made a great step in demonstrating the interest of such materials, but came at the price of poor electronic conduction.

Sun's work,167 gave an excellent resume of the importance of combining polymers with band-gaps of sizes appropriate to solar radiation, the options that the various lamellae, hexagonal and gyroid structures made available through the use of block copolymers, and the importance of the linking group which covalently joined to the two blocks together. This is, in effect, the first paper that laid out a full discussion of the many diverse parameters that must be considered when using block copolymers in OPVs. Subsequent work168 further clarified the need for control over the whole polymer structure. Figure 12(a) shows the designed block copolymer where B is a linking group that covalently bonds the two blocks but interrupts electronic conjugation, and D and A are the donor and acceptor groups, respectively. It was expected that the system would pack as in Figure 12(b) to give the hexagonal structure in Figure 12(c) and hence the device in Figure 12(d). The reasoning behind this vertical profile was that excitonic states could diffuse horizontally and the formed charge move vertically via π-conjugated orbitals. Their actual copolymer studied was based on alkoxy- and sulfonate-PPVs (respectively D and A blocks), bridged by a short diamine unit containing two methylene groups. However, to consistently gain this sort of structure, considerable control over the polymer orientation is required, something which is not always facile and is particularly susceptible to changes in substrates and solvents.119 It was also noted elsewhere, as an intuitive argument, that placing insulating groups right next to the electrodes would considerably diminish charge transfer and that rather, a parallel placement of block copolymers with respect the electrodes might be more appropriate.169 This would also enhance the particularly important polymer-electrode interactions.

Figure 12. (a) The block macromolecule based on donor (D) and acceptor (A) segments with an electronically inert, flexible bridging group (B); (b) the proposed alignment of macromolecules to enhance excitonic formation of charges and their subsequent transfer to electrodes; (c) block copolymer domain formation; and (d) the structures within an OPV. Adapted from Sun et al.180

An alternative approach was taken by Gratt et al.,170 as shown in Figure 13, wherein the incorporation of a triblock copolymer, containing chromophores surrounded by n- and p-type blocks were considered in a horizontal layering of coil-coil polymers carrying pendent groups. It could be argued that for such groups, the specific direction of the polymer backbone is less important, it is reasonably expected that it would be the orientation of the aromatic groups rather than that of the polymer that is of importance.

Figure 13. Schematic of triblock copolymer chains in an OPV, showing incorporation of donor, acceptor and chromophoric materials. Reprinted from Gratt and Cohen,170 Journal of Applied Polymer Science (2004) with permission from Wiley.

It is important to note that the presence of an insulating linker unit was seen as essential to prevent rapid intramolecular charge recombination.171, 121 Indeed, this methodology has been further optimized to give a device with exceptionally high V_oc of around 1.1 V.172 It was also exploited to separate P3HT and polystyrene blocks carrying fullerenes.173 Elsewhere, the size of the flexible linker or “spacer” has been shown to have great influence on the degrees of order attainable by the constituent blocks.174 Nevertheless, very recently Botiz et al. prepared all-conjugated donor–acceptor block copolymers, and it will be of interest to see the effects of such materials in devices.175 Of similar interest is the work by Izuhara and Swager, mentioned in Synthesis section and shown in Figure 8, where a block copolymer based on poly(pyridinium phenylene) demonstrated electron affinities comparable to PCBM even though the block was conjugated with the donor.162

Whatever method chosen, one of the most important parameters remains the size of the domain. Work with blends has shown that domain sizes should be slightly greater than the exciton diffusion length, so as to obtain the most efficient balance between the process of charge formation from excitons, and charge transport with minimal losses by recombination.176 A comparable balance is required when dealing with block copolymers. Watkins et al. released an important paper in 2005, dealing with the Monte Carlo modeling of excitonic diffusion and dissociation, and charge recombination and transport.32 Both blends and model “check-board” structures were studied [other systems, comparable to those seen in Figure 2(c) were not considered]. Two key points were made. First, it was found that the more ordered block copolymer-like structures out-performed more disordered BHJ-like structures by around 150%. Second, the optimum width of each domain was directly related to the product of the efficiencies of exciton collection and charge collection. In the case of the polymers chosen, calculated to have exciton diffusions of 5 nm (see Fig. 14), this worked out to be around 15 nm. In effect, the domain size needs to be exactly of the right size for the efficiency to be maximized. Too small, and charge recombination is favored, too large, and the interfacial density is too low to collect excitons and form charges. Fortunately, this can be managed by carefully adjusting the size of the polymer backbone.

Figure 14. Above, donor/acceptor block copolymer domain-structure used for basis of Monte Carlo calculations; and below, predicted optimum domain widths, found to be a product of maximum efficiencies in exciton and charge collections. Reprinted with permission from Watkins et al.32 Copyright 2005 American Chemical Society.

In the case of rigid conjugated polymers, in general, the width of the domain is close to the length of the polymer. In the case of coil systems, the proportionality is varied by an empirical factor. Furthermore, the width of the domain may be enhanced by highly disperse polymers.157 It is evident that in the case of conjugated polymers, the length of the block can have a particularly pronounced effect on λ_max (the wavelength of light at which the absorbance reaches a maximum), and the processing and electronic behaviors of the material.63 Once more, this is a primary consideration when designing block copolymers for photovoltaic devices.123

It is important to mention the work of Lindner et al.152, 177, 178 where, for the first time, high molecular weight coil-coil block copolymers PvTPA-b-PPerAcr shown in Figure 7(a) were used to prepare devices that clearly demonstrated photovoltaic capabilities greater than blends of the same component homopolymers (Fig. 15). Subsequent studies using related structures indicated the considerable importance of the molar mass of the polymers (coiled systems carrying optoelectronic pendent groups) on device efficiencies.63, 123, 181, 183

Figure 15. Plot of I/V curves indicating power output increase, from 0.007% to 0.07% in going from PvTPA-blend-PPerAcr to PvTPA-b-PPerAcr [see Fig. 7(a) for chemical structure]. Reprinted from Linder et al.178 Angewandte Chemie International Edition (2006) with permission from Wiley.

While much of the above work has relied on using pendent groups on commodity polymers, alternative synthetic strategies have also been created to incorporate acceptor units into the main-chain of the polymer with the eventual aim of better controlling domain widths. Fullerene has been incorporated into a main-chain through a recently discovered atom transfer radical addition polymerization (ATRAP),87 and subsequently reacted with P3HT to yield multiblock copolymers denoted PFDP-b-P3HT [see Fig. 7(d)], which yielded well-defined mesostructures.123

Finally, it should be noted that the materials discussed here are considerably different from the so-called double-cable polymers, principally developed by Cravino using extremely elegant chemistry184 and represented in Figure 16. While these materials might find use in other electronic devices, their use in OPVs will probably be limited by the overly small “domains” formed by each “cable,” lying parallel in such a way that germinate recombinations might occur.

Figure 16. Typical “double-cable” structure.185 Reprinted by permission from Macmillan Publishers Ltd: Nature Materials, Cravino and Sariciftci, copyright 2003.

Compatibilizers

When block copolymers are used as additives to facilitate the blending of immiscible materials, they are described as compatibilizers. Although block copolymers have been used as compatibilizing agents extensively across the literature,186–189 their application as such in organic solar cells remains limited. This is perhaps somewhat surprising as this is an area where block copolymers appear to have their most beneficial impact in OPV technology.40 Block copolymer compatibilizers reduce the interfacial energy between the immiscible components by spanning the heterojunction itself. Generally, one of the segments of the block copolymer will comprise of a material chemically similar to one of the components of the blend whereas the other segment will have chemical similarity to the opposing material. This ensures miscibility between each block copolymer molecule and the constituents of the blend.

Ground-breaking work in this area was first reported by Sivula et al. in 2006.179 In this work, the presence of the diblock copolymer based on pendent P3HT repeat units in one segment and pendent fullerene derivatives in the other, shown in Figure 7(e), drastically improved the thermal stability of P3HT:PCBM device layers (without detriment to the PCE). Pure P3HT:PCBM layers are well known to be thermally unstable as they are processed into a kinetically trapped, nonequilibrium morphology.190 Following heat treatment (annealing), phase segregation is observed where the PCBM moieties aggregate, reducing the interfacial contact area between donor and acceptor. However, annealing is required to increase the mobility of the P3HT domains to improve the electrical characteristics (and thus efficiency) of the device.63 Fréchet and coworkers showed that the addition of 17 wt % of the aforementioned block copolymer [Fig. 7(e)] prevented such thermally-induced macrophase separation and helped to stabilize the performance of the device following extensive thermal treatment (see Fig. 17). Following on from this, in 2009, Fréchet's group demonstrated the same technology for a different blend system based on P3HT with small molecule electron conductors with improved visible light absorption, namely PDI derivatives.180 In this latter work, device performances were shown to increase on addition of 25 wt % of the copolymer poly(3-hexylthiophene)-block-poly(bisimide perylene acrylate) (P3HT-b-PPerAcr), shown in Figure 7(f). However, it should be stated that power conversion efficiencies in this work were considerably lower than those observed in the PCBM-based devices at that time. In the same year, Yang et al.135 synthesized a rod-coil block copolymer denoted P3HT-b-P(S_xA_y)-C₆₀ consisting of P3HT and PCBM, where the PCBM moieties were covalently attached to a flexible polyacrylate-co-polystyrene backbone [Fig. 7(g)]. Only 5 wt % of this block copolymer was required to improve device efficiency of a solar cell based on a P3HT:PCBM blend by approximately 35%. Finally, Lee et al.191 used a similar diblock copolymer, where the PS was replaced with PMMA, and observed its effect on device performance of P3HT:PCBM-based cells during thermal annealing. Again, the presence of the diblock copolymer maintained nanoscale phase separation of the blend thus reducing the detrimental effect of thermal treatment on power conversion efficiencies.

Figure 17. TEM images of a thin film of (a) 1:1 P3HT/PCBM blend before annealing, and (b) after annealing for 1 h at 140 °C. (c) 1:1 P3HT/PCBM + 5 wt % block copolymer after annealing (1 h, 140 °C). (d) 1:1 P3HT/PCBM + 17 wt % block copolymer after annealing (1 h, 140 °C). The white scale bars represent 2 microns and the dark areas indicate fullerene-rich regions. (e) shows the average power conversion efficiencies versus annealing time with (circles) and without (squares) block copolymer compatibilizer. Annealing was performed before and after Al deposition (open and filled symbols, respectively). Reprinted from Sivula et al.179 Advanced Materials (2006) with permission from Wiley.

Although there are only a handful of examples reporting the use of block copolymer compatibilizers in organic solar cells, it is highly likely that this approach will have exponential growth in the coming years, due to its drastic effects at extremely low additive contents.

Templating

There are many examples across the literature where researchers have investigated the use of block copolymers as templating agents for use in OPVs. In this section, we briefly describe the strategy and discuss select instances to provide an overview of the scope of this burgeoning technology. For further examples of block copolymer templating in OPVs, the reader is directed elsewhere.192–199

As discussed earlier, block copolymers have a rich phase diagram enabling a variety of complex structures to be assembled, which primarily depend on the volume fraction of the blocks and their molecular weights. Using the aforementioned well developed design rules it is relatively straightforward to produce lamellar, cylindrical and spherical architectures with pre-programmed domain sizes. After ordering a block copolymer layer it is possible to selectively remove one of the components (generally by degradation) to leave a mold where the voids in the film can be backfilled with another material. Backfilling can be carried out using either dip coating or electrochemical deposition. Some of the more complex and interconnected block copolymer structures, such as the gyroid phase, occupy only a very small region of the phase diagram and complications due to the rod-like nature of some of the blocks make accessing these morphologies very difficult. This makes it necessary to use more conventional coil-coil block copolymer templates to attain them. The alternative approach involves synthesizing a large number of block copolymers and mapping out the entire phase diagram, which is clearly a more laborious route, but has significant academic merit nonetheless. Crossland et al.200 (see Fig. 18) utilized a diblock copolymer that forms a highly interconnected double gyroid structure. After ordering the layer via thermal annealing, one component was selectively etched away and titanium (IV) oxide was then electrochemically deposited within the voids. The remaining block copolymer component was thermally degraded and removed and the layer was subsequently backfilled with dye sensitizer.

Figure 18. The gyroid templated route where (a) a poly(4-fluorostyrene)-block-poly(D,L-lactide) (PFS-b-PLLA) diblock produces a highly interconnected network structure. After PLLA removal (b), the PFS is electrochemically replicated (c) and the empty space is backfilled with dye sensitizer (d). Reprinted with permission from Crossland et al.200 Copyright 2009 American Chemical Society.

This approach shows impressive control over structure and ordering, however, the device efficiencies only reached 1.7%. For a complex multiple stage process like this to be viable, the device efficiencies would have to be considerably improved when compared to existing one/two stage polymer blend devices. Nonetheless, this is an elegant processing approach which demonstrates the technology and will inspire progress in this area.

A series of rod-coil poly(3-alkylthiophene)-block-poly(L- lactide) (P3AT-b-PLLA) block copolymers with alkyl chains of either 6 or 12 carbons have been synthesized and studied to measure the effect of the PLLA block on the overall optical properties and on the P3AT crystallinity.146 After ordering (via thermal annealing), these materials displayed absorption spectra similar to the P3AT homopolymer, having the same distinct vibronic structure. The PLLA was degraded by weak alkali solution to leave the P3AT component, which retained the same optical absorption properties while adopting a weakly ordered net-like morphology with pores 35 nm in diameter. Another study utilized a P3HT diblock with a degradable PLLA segment to form a vertically aligned lamellar structure with a period of ∼16 nm, as shown in Figure 19.201 The etched PLLA region was backfilled this time by dip coating the film in an aqueous C₆₀ solution, to fill the voids between the P3HT regions. Evidence for quenching in the layer was measured by photoluminescence although no actual OPV devices were fabricated. Boer et al.137 demonstrated how a drop cast layer of a poly(p-phenylene vinylene)-block-polystyrene system will order into a very periodic structure.137 Any system that rapidly orders lends itself to templating, provided that the nonconjugated component can be removed successfully.

Figure 19. Schematic to show the use of a degradable PLLA domain in a P3HT-b-PLLA diblock (a) to form vertical lamellae domains. The sequential processing steps comprise of ordering (b), removal of PLLA (c) and backfilling with C₆₀ (d). The random structure without the ordering and processing steps is shown in (e). Reprinted with permission from Botiz and Darling.201 Copyright 2009 American Chemical Society.

Ordering of fibril structures has been demonstrated for a PMMA homopolymer-blend-(P3HT-block-PMMA) block copoly mer blend system, where PMMA represents poly(methyl methacrylate).202 In this work, Choi et al. added PMMA homopolymer to explore the phase diagram by altering the ratio of homopolymer to block copolymer. Addition of homopolymer allowed the fibrils to reorient to give highly ordered fibril structures over relatively large volumes. Subsequently, the addition of PCBM did not disrupt the ordering, showing that the PCBM was evenly distributed and gave rise to quenching to demonstrate charge transfer. However, given the high content of dielectric PMMA, the authors did not expect good efficiencies with this material.

One of the main drawbacks with the templating approach is the processing time and effort required to order the layer, remove one phase and then backfill. Other difficulties with this approach include the degree of infiltration, as the backfilled component must be distributed throughout the entire thickness of the film and not just at the surface. This problem seems to have been resolved, however, by the use of a gyroid morphology which allows for broken pathways to be present, providing numerous routes throughout the structure.196, 200 Intuitively, the coupling of the backfilled component with the interface of the other component is another consideration; as there must be intimate contact between the two interfaces to ensure charge separation and avoid exciton losses.

PROCESSING

In the past 10 to 15 years there has been intense research focused on the optimization of polymer-based solar cells. The approach to improve device efficiency has involved varying numerous parameters such as the choice of donor and acceptor, donor:acceptor ratio, choice of casting solvent(s), and annealing treatments. In contrast to this, one of the main advantages of using block copolymers as solar cell materials is that they would not require a combinatorial screening approach to fully optimize their device efficiencies. Careful planning and judicious choice of the structure at the outset would help to target a particular donor–acceptor domain size, the type of morphology and the interfacial width between the two components. Block copolymers will proceed to the microphase-separated equilibrium structure regardless of the casting solvent (as long as the solvent is nonselective) and numerous other processing parameters. In this section we highlight a number of processing methods capable of ordering block copolymers that might be attractive routes for block copolymer-based OPVs. For further in-depth discussion into the directed ordering of block copolymers, the reader is directed elsewhere.203

The drive to use block copolymers in organic solar cells is mainly due to their equilibrium structure on a well-defined length scale (as discussed in Microphase Separation section). Ideally, the processing of the active OPV layer needs to be simple and rapid, so that it is compatible with existing roll-to-roll (R2R) technology akin to the manufacture of conventional polymer films that have also recently been demonstrated for solar cell blend manufacture.163, 204, 205

Conventional routes to deposit thin uniform films on substrates over large areas involve spin coating or doctor blading. The main drawback with these deposition methods is that they do not produce well-ordered block copolymer states (particularly the spin-coating approach) as they are highly nonequilibrium processes where rapid solvent removal quenches the polymer morphology without allowing the system time to reach the lowest energy equilibrium state. It is, therefore, essential to postprocess after deposition of the block copolymer thin film, as shown in Figure 20, where solvent annealing was used to order the block copolymer.143 Müller et al.206 synthesized a diblock copolymer of poly(3-hexylthiophene)-block-polyethylene (P3HT-b-PE) and observed that this system has good conductive properties alongside mechanical toughness, both qualities that are desirable for flexible solar cells. Although not strictly concerned with making OPVs from this material, it shows the possibilities of a double crystalline diblock system. Their study looked at the effect of processing, as PE is a commodity polymer that is well studied and various processing techniques have been used and optimized for PE. The novelty was that for low P3HT content in the diblock, charge transport was comparable to pure P3HT. They suggested that as the P3HT crystallizes first, the spherullitic PE structure does not disrupt the P3HT domains, and so maintains a co-continuous path for charges.

Figure 20. Thin films of the copolymer P3HT-b-PPDA [structure shown in Fig. 7(h)] imaged using SFM [(a) and (c)] and TEM [(b) and (d)]. The sample shown in (a) and (b) was annealed for 2 h in a chloroform/toluene vapor, whereas slow solvent evaporation over a day gave the samples seen in images (c) and (d). Reprinted with permission from Zhang et al.143 Copyright 2009 American Chemical Society.

For block copolymers, a number of factors will affect the ordering and its rate, such as χN, chain mobility, viscosity and diffusivity above the glass transition temperature. Solvent or thermal annealing will allow an initially mixed block copolymer system to microphase separate to minimize energy. An important aspect when making block copolymer thin films is that one of the block copolymer components may have preferential attraction to the substrate and potentially act as a hole or an electron barrier layer, depending on the device architecture. This arises due to the differences in surface energy between the components of the block copolymer, as mentioned earlier (see As Single and Multicomponent Layers section) in the context of massive improvements in device efficiency from 0.03% to 1.22% by changing device architecture from conventional to inverted due to the preferential wetting of P4VP on the PEDOT-compl-PSS layer.166 To avoid this, the substrate can be treated either by plasma polymerization to make a neutral surface for the block copolymer or by graphoepitaxy.207 Graphoepitaxy is a technique where a substrate surface relief is used to guide the structure and provide ordering of the overlying block copolymer film. However, this technique still requires a long thermal annealing stage for the block copolymer layer to conform to the underlying surface relief pattern. Relatively recently, Welander et al.208 used high temperature annealing (∼250 °C) alongside a chemically nanopatterned surface to achieve rapid ordering (∼1–5 min) of a lamellar-forming PS-block-PMMA diblock that has in the past taken many hours.

Another approach is to use shear ordering, a well-established means of aligning block copolymers in the bulk209, 210 that can also be used to provide order in thin films. Angelescu et al.211 demonstrated how an initially disordered thin film of cylinder-forming PS-block-PEP (where PEP is poly(ethylene-alt-propylene)) could be aligned using unidirectional shear applied via a poly(dimethylsiloxane) (PDMS) pad. Their straightforward method imparted a shear stress on the film and produced highly ordered cylinders with a defect density an order of magnitude lower than a thermally annealed sample. This shear ordering technique can be used over large areas and, by managing the applied lateral forces, the degree of ordering can be controlled. Such a technique is particularly pertinent for organic photovoltaic technology where the presence of so-called nanopillars running perpendicular to the electrodes would be ideal for charge separation and percolation. In other work, Albrecht et al.212 showed how an electric field can be used to induce order into an asymmetric PS-block-PMMA diblock copolymer, using electric field strengths of approximately 30 V μm⁻¹. By applying the electric field while heating the layer above its glass transition temperature, the PS cylinders were made to orient perpendicular to the substrate. The main drawback with this approach is that it still requires many hours to produce ordered films, the extent of which being dependent on the annealing temperature, which is clearly a constraint for an industrially viable process.

An interesting development in this area is the use of microwave-assisted ordering. This is essentially solvent annealing of the layer at an elevated temperature as the microwave radiation induces commensurate heating of the block copolymer layer.213 The advantage of this approach is that the annealing time only needs to be on the order of 100 seconds and has similar defect densities to samples that have been thermally annealed for 36 hours. Interestingly, significant ordering is also possible on the 2 to 5 second timescale and, as OPV materials do not need to be addressed like hard drives or other data storage media (where exact registry of neighboring domains is required), this short annealing time would be adequate for the ordering block copolymer OPVs. Figure 21 shows the quality of rapid ordering that is possible using this approach.

Figure 21. Directing order via microwave annealing in the presence of THF as observed by SEM. (a) and (b) show platinum templated lines via microwave annealing of PS-block-P2VP for 60 seconds at 130 °C. (a) shows the polymer sample on an unpatterned substrate, whereas the substrate in (b) had a circular topographical feature. (c) shows another block copolymer on unpatterned silicon, whereas (d) shows the sample with parallel SiO₂ lines guiding the block copolymer ordering, both with 180 seconds microwave annealing time. Reprinted with permission from Zhang et al.213 Copyright 2010 American Chemical Society.

Rapid thermal annealing strategies have been applied to a conjugated P3HT diblock system with C₆₀ incorporated onto the main chain. Scanning force microscopy phase images showed a high degree of order after annealing for a few minutes. In this example, 5 min at 220 °C was sufficient to produce nanoscale fibrils with domain thicknesses of approximately 19 nm.123 However, it has been shown that the annealing step may not be required if the polymer is able to crystallize on a short timescale, thus creating pathways for charge transport. Dante et al.173 prepared a rod-coil-coil triblock copolymer based on P3HT, PS, and PS carrying pendent C₆₀ molecules that when spin coated gave well ordered P3HT fibrils 250 nm long and 11 nm wide. They also demonstrated quenching of the photoluminescence showing that there is intimate contact between the P3HT and the C₆₀ domains.

Any processing or ordering procedure used to fabricate solar cells will need to be fast and produce sufficient order that makes block copolymer solar cells that are superior to polymer blend systems. A number of approaches are possible as discussed herein but their suitability will depend on how they affect a given polymer system. In particular, oxidative damage or other photodegradation of the conjugated polymers must be avoided.74

CURRENT BEST DEVICES

Table 1 highlights the characteristics and structural order of some of the block copolymer solar cell materials published across the literature. In compiling this list, which is by no means comprehensive, we note that it is often difficult to compare all of the many systems that have been reported. Occasionally, researchers do not fabricate devices out of their materials making it difficult to judge a given system's merits. However, the selected examples reflect the diversity and varied approaches to the design and implementation of block copolymer-based solar cell devices and architectures. To understand the table and some of the parameters used to characterize solar cell devices refer to the brief summary of how the design and intrinsic mechanisms of solar cell affect their properties provided in Introduction section. More detailed explanations can be found elsewhere.3, 23, 215, 216

Table 1. Device Characteristics for Block Copolymer-Based Solar Cell Devices

The following discussion covers each of the block copolymer devices featured in Table 1, highlighting the various aspects considered to be important in an attempt to inform others in the design of future block copolymer systems.

Although the final device was not composed of a block copolymer active material,200 the aforementioned work of Crossland et al. showed that a block copolymer can be used in a templating approach to create a device with a PCE of 1.7%. This method, however, required four separate processing stages and would need to be improved, both in terms of PCE and processing, before this technology can be realized on an industrial platform.

The group of Thelakkat181 compared two block copolymers having the same composition but with one approximately half the molecular weight of the other. They used the earlier reported poly(3-hexylthiophene)-block-poly(perylene bisimide acrylate) copolymer [Fig. 7(i)]182 and comparable to those of the groups Fréchet [Fig. 7(f)],180 and of Russell and Emrick [Fig. 7(h)]143 but with different chain-ends and linking groups. They found that the higher molecular weight polymer had improved charge transport properties, which they correlated with increases in the hole carrier mobility due to an increase in crystallite domain sizes. However, in this example, devices gave at best a PCE equal to 0.2%.

Segalman et al.132 later prepared a comparable poly(3-hexylthiophene)-block-poly(n-butyl acrylate-stat-acrylate perylene), interestingly via click chemistry. They examined the link between order and the effect on device properties using AFM and transmission electron microscopy (TEM), utilizing solvent annealing to order their spin cast films. The domain sizes of their block copolymer were ∼ 25 nm (depending on the processing parameters) and gave nanostructured cylinders parallel to the substrate. Clearly, an ideal morphology would comprise of such cylinders being perpendicular to the electrodes. They also studied the effects of adding plasticizing butyl acrylate to aid the ordering of their layer. The best device efficiency was 0.2% with a low J_sc, highlighting limitations in light absorption, the efficiency of generating charges and their ultimate collection at the electrodes.

As mentioned above, the block copolymer PvTPA-b-PPerAcr shown in Figure 7(a) was synthesized along with the respective homopolymers.178 This enabled the comparison of blend and block copolymer devices for the same composition of the two polymers. Comparing the morphology using cross sectional TEM they found that the blend formed micron-sized domains whereas the block copolymer gave ribbon-like domains (∼100 nm in length and 10 to 20 nm wide) aligned parallel to the substrate. The block copolymer had an order of magnitude better PCE than the comparable blend (see Fig. 15). This work serves to show the difference between systems comprising of the same fundamental building blocks. However it would be even more beneficial to future progress to compare any improvements in efficiency with comparable sizes of morphology, by perhaps using a solvent that gave a finer length scale for the blend morphology or quenching the blend system to kinetically trap the morphology.

The aforementioned P3HT-b-PPDA prepared by Zhang et al.,143 shown in Figure 7(h), was characterized as thin films using AFM and TEM. This showed ∼ 25 nm wide fibrils that were microns in length following solvent annealing. The device efficiencies improved from an initial PCE value of 0.11% for the as-cast film to 0.49% for the thermally annealed devices. Miyanishi and coworkers124 synthesized and used the block copolymer carrying fullerene moieties, P3HT-b-(PCBM-P3HT) shown in Figure 7(j) and found that after annealing at 130 °C their devices had a PCE of 1.7%. Importantly, they demonstrated one of the major advantages of utilizing block copolymers as solar cell materials. They compared a blend and a block copolymer device under extended annealing conditions, deemed equivalent to ageing the layers. The blend efficiency declined from an initial value of 3% to ∼ 0.5% due to the mass transport of PCBM to the cathode interface and the formation of needle-like PCBM crystals. Conversely, the block copolymer layer efficiency remained constant at the value achieved after the initial thermal annealing used to order the nanoscale structure. It is this long term stability that may provide the impetus to switch from blend to block copolymer solar cells, provided that the problem with the inherent lack of efficiency associated with block copolymer systems is overcome.

In work by Bu et al. (shown in Fig. 22),214 an oligo(fluorene)-block-oligo(bithiophene) material was used to try and overcome some of the problems of chain rigidity, ordering and dispersity effects associated with rod-like block copolymers to produce alternating donor–acceptor lamellae. TEM and selected area diffraction demonstrated that the oligomers formed layers perpendicular to the substrate with a period of 5 nm to 10 nm, depending on the oligomer architecture. Their devices had efficiencies of up to 1.5%, and a FF of 38% suggesting that charge movement to the electrodes is not as superior as might be expected for this perpendicular architecture. Furthermore, the absorption spectrum was not ideally matched to the AM 1.5 solar spectrum. However, this system is one of the best single component OPV systems reported to date.

Figure 22. TEM images and selected area diffraction images for a lamellar-forming oligo(fluorene)-block-oligo(bithiophene) showing the improvement in order for solvent annealed (a, b and c) and thermally annealed (d, e and f) samples. The absorbance in (g) is given for solution (sol), pristine (pris), solvent annealed (SA) and thermally annealed (TA) samples. A cartoon figure in (h) shows the structure of the lamellar nanostructures in the films. Reprinted with permission from Bu et al.214 Copyright 2009 American Chemical Society.

Finally, another approach is to use a mixture of a block copolymer and PCBM, where the block copolymer provides the optimal morphological length scale with long term stability. However, this may not be as stable as a fully conjugated donor–acceptor block copolymer system due to the migration and movement of PCBM through the polymer film. For example, recent work has shown that the complete diffusion of PCBM through amorphous P3HT (100-nm thick film) takes place within ∼5 s at 150 °C.217 The work of Gu et al.218 provides an example of a block copolymer and PCBM system. Here, they synthesized a P3HT-block-PS copolymer and added PCBM to make devices that had a maximum optimized device efficiency of 1.93% after thermally annealing the layer. Conversely, initial as-cast devices had efficiencies of only 0.4%. Annealing was shown to improve the nanoscale phase separation thus leading to an increase in the overall efficiency of the device.

CONCLUSIONS

It is clear, both from theoretical and empirical results, that block copolymers offer a rational and pragmatic route to preparing highly efficient OPVs. However, their current lag behind BHJs will have to be solved, or at least reduced, before they can be taken seriously in the wider academic and industrial communities. One of the most important aspects of block copolymers, and one that is a considerable strength as it allows for great variations, is the high number of parameters that must be addressed before they might be used in efficient devices. In terms of their structures, one has to consider chain lengths which impact on their optoelectronic capabilities (band-positions, λ_max and electron confinement), their electronic properties (exciton “capturing”) and charge transfer, their processing (solubilities), and the mesoscale morphologies that the polymers will take up. The types of polymers that are used will impact again not only on their absorptions and electronic band interactions, but also on the electronic, intra-domain crystallization and aggregation, and the final morphological take-up (through influencing χN). Linking groups are also important, not only in terms of their electronic effects on charge separation and recombination, but also in permitting flexibility and the self-assembly of the donor and acceptor blocks. Furthermore, the synthetic pathways will most likely have to be optimized with respect to their work-up in industrial environments and any traces of impurities which might remain in the final device. The optimum morphology, with respect to the character of the polymers, their orientation with respect to the electrodes (and the overall device architecture, be it “normal” or “inverted”), optical effects, and the optimum intra- and inter-polymer charge-carrier pathways will also have to be considered. Finally, the processing conditions, such as solvents, substrate-effects, casting rates, annealing steps and so on will have to be detailed. Beyond all this, an often heard argument against the use of block copolymers in OPVs is their possible cost. Experience has shown though that, in any industry, if a system works well enough, then there will be commercial interest. Once this starts, then economies of scale can come into play. Given the underlying theoretical foundations, and the enormous number of ways in which block copolymers can be modified, it could be reasonably said that block copolymers will probably play, in some fashion or other, a major role in the future development of viable OPVs.

Acknowledgements

AJP is funded by the EPSRC via grant EP/E046215 (Soft Nanotechnology).

Biographical Information

Paul D. Topham

received his Ph.D. in 2006 from the University of Sheffield under the supervision of Professor Tony Ryan, working on molecular machines. After a post doctoral position with Professor Steve Armes, working on polymer-peptide conjugates, Dr. Topham was appointed as a Lecturer in Chemistry at Aston University in 2008. His current research interests include controlled polymerisation, click chemistry, polymer-peptide conjugates and electrospinning polymer nanofibres for a wide range of applications.

Biographical Information

Andrew J. Parnell

obtained his Ph.D. on the study of responsive nanosystems in 2007 from the University of Sheffield, under the supervision of Professor Richard Jones FRS. His postdoctoral research career began under Dr. Patrick Fairclough before his return to the group of Prof. Jones, in both positions studying block copolymer self-assembly. Dr. Parnell's current research involves understanding the link between structure property relationships for light emitting polymer devices, organic solar cells and structural color.

Biographical Information

Roger C. Hiorns

received his Ph.D.s from the Universities of Kent and Montpellier II in 1999. After postdoctorial studies at the Université de Pau in the group of Dr. B. François and Dr. J. François, and then at the ENSCBP in Bordeaux with Professor Henri Cramail and Dr Eric Cloutet, he was appointed Chargé de Recherche (CNRS). He is currently at the Université de Pau et des Pays de l'Adour, France. His research centers on synthetic routes to fullerene-based and multiblock copolymers.