Many advances in organic photovoltaic efficiency are not yet fully understood and new insight into structure-property relationships is required to push this technology into broad commercial use. The aim of this article is not to comprehensively review recent work, but to provide commentary on recent successes and forecast where researchers should look to enhance the efficiency of photovoltaics. By lowering the LUMO level, utilizing electron-withdrawing substituents advantageously, and employing appropriate side chains on donor polymers, researchers can elucidate further aspects of polymer-PCBM interactions while ultimately developing materials that will push past 10% efficiency.
Featuring significantly reduced cost on both materials and fabrication when compared with the market dominant crystalline Si solar cells, organic solar cells have been touted as a serious contender to lead the next generation of solar cells. Thus, the field of organic solar cells has attracted a tremendous amount of research activity. A simple search of the Web of KnowledgeSM using the key words “organic solar cells” returned over 8000 results! As shown in Figure 1a, the number of publications has been rapidly increasing in the past 10 years, in particular within the past 5 years (Figure 1b), which clearly indicates the rapid growth of this research field.
Among all “organic”-based solar cells, polymer solar cells, in particular polymer/fullerene-based bulk heterojunction (BHJ) solar cells,1 are arguably one of the hottest research fields.2, 3 By blending the electron-donating semiconductor (DONOR, e.g., polymers) and an electron-accepting semiconductor (ACCEPTOR, e.g., fullerenes) in bulk, the BHJ offers some unique advantages and functions as follows (Figure 2). First, the light absorption by organic semiconductors only produces excitons (tightly bound electron-hole pairs), which need to travel to the DONOR–ACCEPTOR interface to separate into energy-carrying charges. However, these excitons usually have a very short lifespan and a similarly short diffusion distance (≈10 nm). Thus, the minimized travel distance to the DONOR–ACCEPTOR interface rendered by the BHJ configuration is beneficial for efficient exciton dissociation. Second, the BHJ maximizes the interfacial area between the DONOR and the ACCEPTOR, and allows one to employ films of thicknesses much larger (typically 100–200 nm) than the exciton diffusion length (≈10 nm). A thick film can absorb more photons, thus more excitons can split into usable charges. Finally, the interpenetrating network of the BHJ offers charge transport pathways to assist the charge collection at the electrodes.
Empowered by the synergistic efforts among chemists, physicists, and engineers, the power conversion efficiency of BHJ solar cells has been steadily increasing (Figure 3). From the materials' perspective, poly(phenylene vinylene)s (PPV) dominated the research field in the 1990s, such as poly[2-methoxy-5-(2′-ethylhexyloxy)-1,4-phenylenevinylene] (MEH–PPV) and (poly[2-methoxy-5-(3,7-dimethyloctyloxy)]-1,4-phenylenevinylene (MDMO–PPV). Through the application of chlorinated solvents to tune the morphology of the active layer (i.e., the blend of polymer and fullerene derivatives), up to 3.3% power conversion efficiencies were achieved in PPV-based BHJ solar cells with PC61BM as the acceptor material ([6,6]-phenyl C61-butyric acid methyl ester, a soluble version of the original C60).4, 5 The next efficiency milestone was achieved by poly(3-hexylthiophene) (P3HT), which has been extensively studied since the early 2000s.6–8 Again, the careful control of the morphology of the BHJ blend of P3HT:PC61BM ultimately resulted in ≈5% efficiency.3, 8–10 However, with relatively large band gaps, both PPVs and P3HT cannot absorb enough light, severely limiting further efficiency improvement. Therefore researchers have pursued novel polymers of lower band gaps, in order to harvest more light thereby potentially attaining higher efficiency. In the past few years, the field has witnessed the development of several new polymers, with a few achieving 7–8% efficiency in typical BHJ devices with fullerenes as the acceptor.11–22 Very recently, a record high efficiency of over 10% was reportedly achieved by Mitsubishi.23, 24 All these accomplishments are a testament to the significant progress achieved by the organic photovoltaic (OPV) research community.
In response to the rapid growth of this exciting research area, a number of excellent reviews have been dedicated to the topic of polymer solar cells. These reviews have covered various aspects of this interdisciplinary research field, such as design of polymers,26–28 device physics,29, 30 physical chemistry,31, 32 morphology control,33–39 and stability/economics.40, 41 Rather than contributing another comprehensive review, we attempt to direct the readers' attention to the latest advances in the design of new polymeric materials for BHJ solar cells. We will focus on the outstanding issues in the molecular design of conjugated polymers that warrant further research activities, such as (1) lowering the lowest unoccupied molecular orbital (LUMO) energy level and enhancing the external quantum efficiency (EQE), as well as advantageously utilizing (2) electron-withdrawing substituents and (3) side chains. For each section, we will begin by discussing a few selected molecular systems, so as to introduce empirical guidelines for future design. We will then recommend additional research directions not yet fully explored. In doing so, we aim to further inspire creative molecular designs from the research community, in order to reach even higher efficiencies.
2. Lower LUMO Energy Level and Higher EQE
Excitons in organic semiconductors typically have a binding energy between 0.1–1.0 eV,42, 43 and thus photovoltaic cells employing organic semiconductors (typically p-type) require an additional semiconductor (typically PC61BM as the n-type) with a lower LUMO energy level to split these Frenkle excitons. However, even though conventional wisdom quotes a 0.3 eV driving force required to dissociate an exciton from a conjugated polymer, the vast majority of conjugated polymers developed in the past five years have a LUMOpolymer − LUMOPCBM gap (ΔEED) much greater than the 0.3 eV required. For example, regioregular P3HT has a measured ΔEED of 1.1 eV,44 which gives a 0.8 eV excess energy that is wasted when the excited electron is transferred to PCBM (Figure 4).45 Therefore the primary method for increasing the performance of the conjugated polymer is to decrease ΔEED to as close to 0.3 eV as possible.46 This would help achieve both a small band gap and a low highest occupied molecular orbital (HOMO) energy level, in order to get both a high short-circuit current (Jsc) and a high open-circuit voltage (Voc).
2.1. Current Status on the LUMO Level Engineering
Table 1 shows the top eight polymers which have achieved power conversion efficiencies above 7%, and the corresponding ΔEED of each polymer. The polymers with the lowest ΔEED of 0.4 eV are entries 3 and 4, employing the electron-deficient thieno[3,4-c]pyrrole-4,6-dione (TPD) monomer. TPD has been a very popular monomer recently in the literature, with three groups recently reporting polymer cells over 7% efficiency with this particular monomer unit,17–20 among other high-performing ones.47, 48 The measured electrochemical LUMO for TPD materials is typically around –3.9 eV, which is the lowest electrochemical LUMO ever reported for a material with over 7% efficiency. Its widespread success is likely due to the low ΔEED for this class of materials. However, the EQE values for this family of polymers remain below 70%, therefore, additional work is required to optimize the other factors which govern photovoltaic performance that have allowed other materials with larger ΔEED values to reach EQE values greater than 70%.
Table 1. Top eight polymer solar cells over 7% and their photovoltaic properties.
a)All HOMO/LUMO levels use Fc/Fc+ as –4.8 eV from vacuum. PCBM = –4.3 eV.
2.2. Promising Electron-Deficient Structural Units
In order for BHJ photovoltaic cells to reach 10% or higher efficiency with PC61BM, the LUMO of conjugated polymers must be reduced further still to at least –4.0 eV while maintaining a high EQE value. Therefore, in order to synthesize polymers with exceptionally low LUMO energy levels, new easily reduced aromatic moieties which can be readily included into conjugated polymers are required.
The most common method for synthesizing low-bandgap copolymers is the intramolecular charge transfer (ICT) approach,49, 50 in which the HOMO and LUMO energy levels are determined by different monomers, allowing the synthetic chemist to independently control both energy levels. The most widely investigated ICT LUMO reducing materials are based upon 2,1,3-benzothiadiazole (BT). One such material has reached IQE values of near 100% with a LUMO energy level of –3.6 eV.15 Recent research has focused on designing aromatic moieties, which are more electron-deficient than BT, by either adding electron-withdrawing groups, pyridinal nitrogens, or additional electron-deficient rings to the benzothiadiazole core.
Pyridazine-based monomers are one promising yet unexplored family of electron-deficient heterocycles that have measured LUMO energy levels between –3.88 and –4.15 eV (Figure 5). Gendron and co-workers have led initial studies into these heterocycles as acceptors for conjugated polymers, showing significant results.51 The key drawback for these reported materials is the low molecular weight, likely due to inhibition of the palladium catalytic cycle during polymerization. This drawback has kept performance below 1% efficiency for this class of materials. However, the promising LUMO level of these materials warrants further study into methods, which could deliver high-molecular-weight polymers based upon pyridazine electron acceptors (Table 2).
Table 2. Series of promising heterocycles, which have measured LUMO energy levels of –3.9 or lower that have not reached greater than 6% efficiency.
Monomers based upon indigo dye are another class of electron-deficient heterocycles which have the potential to provide low LUMO levels. Initial investigations by Reynolds and co-workers53 have developed isoindigo as an electron-deficient moiety, yielding p-type chromophores with LUMO energy levels as low as –3.9 eV. When copolymerized in a typical ICT fashion through Stille coupling polycondensation, these systems yield power conversion efficiencies of over 4.0%. These initial results could likely be improved upon,57–60 and indigo- and isoindigo-based systems are especially intriguing because of their ability to attach alkyl chains to the LUMO reducing unit.
2.3. The Issue of Low Absorption Coefficient
One major drawback of using exceptionally electron-deficient benzothiadiazoles and other electron-deficient acceptors for use in ICT copolymers is that the LUMO and HOMO are quite often located on different parts of the polymer, rather than delocalized along the polymer chain. This leads to relatively weak absorption coefficients, since excitation from the HOMO to the LUMO becomes quantum mechanically disallowed. An extreme example of this shortcoming is the case of polymers synthesized from cyclopenta[2,1-b:3,4-b′]dithiophen-4-one (CPD)61 shown in Figure 6.
CPD-based systems such as the polymers and small molecules shown in Figure 7 exhibit exceptionally low LUMO levels, with malonitrile condensation derivatives such as (3) reaching LUMO levels below –4.2 eV. The CPD monomer is so easily reduced because the unreduced form is a 13 electron ring system, one electron short of the 14 required to fulfill Hückel's rule. However, even though polymers and small molecules synthesized with CPD-based systems possess very low electrochemical band gaps, the optical absorption in the low energy portion of the spectrum is typically very poor.62
Similar poor absorption coefficients in the infrared portion of the absorption spectrum are seen in the case of benzo[1,2-c;4,5-c′] bis[1,2,5]thiadiazole-based copolymer systems as well, due to the same issues.63, 64 Therefore, when designing new acceptors for ICT polymers, emphasis needs to be placed on delocalizing the LUMO along the polymer backbone, rather than localizing it on only a few atoms. Otherwise, low absorption coefficients will result.
Thus, while many successful electron-deficient monomers have been synthesized, there has still not been one comonomer, which allows for an optimal LUMO and EQE values above 70–80%. The next generation of LUMO-reducing monomers must be designed with optimal LUMOs, high absorption coefficients, and structures that promote fast charge extraction from the active layer to achieve maximum performance.
3. Influence of External Substituents
A growing trend has been to incorporate electron- withdrawing substituents into the polymer structure, which in many cases have led to dramatic enhancements in solar cell performance.11, 21, 22, 65 It has already been demonstrated that they can effectively lower the HOMO and LUMO levels.66 However, researchers have yet to determine why these substituents, especially the fluorine atom, seem to have a good effect on the hole mobility, morphology, and charge dissociation of the polymer. The following section will categorize examples based on substituent location (on the electron-deficient acceptor moiety or the electron-rich donor moiety) and attempt to survey how photovoltaic properties are impacted.
3.1. Substitution on the Electron-Deficient Acceptor Moiety
Polymer backbones substituted with fluorines on the most electron-deficient unit have received widespread attention for their exceptional performance in solar cells. Three of the top polymers achieving over 7% efficiency contain the benzodithiophene (BnDT) unit copolymerized with a fluorinated acceptor moiety such as thienothiophene (TT),11 benzotriazole (TAZ),22 and benzothiadiazole (BT).21 Table 3 lists the photovoltaic properties compared with their nonfluorinated counterparts, and as can be seen, fluorinating the acceptor moiety seems to lead to better photovoltaic properties all around.
Table 3. Photovoltaic properties of high-performing fluorinated polymers and their nonfluorinated counterparts.
The PTB polymer series was the first to draw attention to incorporating fluorine into DONOR polymers and thus will be the main focus in this section because many studies have already been conducted on this series. Fluorine was originally introduced to the 3-position of the TT moiety as a second electron-withdrawing group (the first being the ester alkyl group) to further lower the HOMO level and therefore enhance Voc.65, 68 Studies on PTB have shown that fluorine only lowers the HOMO level by –0.15 eV (PTB9 vs. PTB7) while the Voc improves from only 0.60 to 0.74 V.11, 67
In an attempt to further optimize the HOMO level of PTB polymers, attention was turned toward other electron- withdrawing substituents. Table 4 shows the various methods in which TT has been modified. Interestingly, when TT was substituted with only a ketone (entry 3), the HOMO level was brought down to –5.12 eV, indicating that a ketone has a comparable electronic impact on PTB as do an ester and fluorine combined.69 When a ketone and fluorine were used in conjunction along with an alkyl chain on the BnDT unit (entry 4), the HOMO level significantly lowered to –5.34 eV.70 However, further attempts to use the even more electron-withdrawing sulfonyl again yielded a HOMO level of only –5.12 eV (entry 5).71 When Ikai and co-workers72 employed phenyl ester pendants 4-fluorophenyl and 4-(trifluoromethyl)phenyl, deep HOMO levels of –5.39 eV and –5.60 eV were observed (entries 6 and 7). However, polymers exhibited rather low mobilities (2.8 × 10−5 and 1.4 × 10−5 cm2 V−1 s−1, respectively), most likely due to the lack of a side chain on TT and the extremely bulky 2-octyldodecyloxy solubilizing chain that was needed on the BnDT unit.72
Table 4. Various methods of modifying thienothiophene and resulting photovoltaic properties.
*Please see respective references for processing conditions and fullerene material used.
In an effort to remove reliance on external substituents, a nitrogen atom was introduced into TT (entry 8), thereby changing the unit to the more electron-deficient thienothiazole (TTz), which can also stabilize its quinoid form.73 Initial results for PBnDT-TTz showed a higher efficiency of 2.5% compared with its direct TT analog, but the HOMO level of this TTz-based polymer was still not quite low enough. Just recently, Yu and co-workers67 reported selenium-based derivatives of their PTB series. The resulting –5.05 eV HOMO level of PBSe1 (entry 9) was similarly high as its sulfur-based analog (entry 1).
Of the various electron-withdrawing groups used, fluorine appears to be one of the most promising because it not only lowers the HOMO level but appears to improve morphology. PTB7, which achieved a previously record-breaking 7.4% efficiency,11 has demonstrated a very favorable morphology. The “zig-zag” shape of PTB's backbone is credited with being responsible for its face-on orientation, which allows for maximal contact with the electrode.74 Furthermore, a grazing incidence wide-angle X-ray scattering (GIWAXS) study proposes that within the active layer, a hierarchy exists ranging from PTB7 nanocrystallites > interpenetrating regions of polymer and fullerene > PCBM nanocrystallites (Figure 8).75 The PTB7 crystalline aggregates are believed to be responsible for the high photocurrent observed because its crystallinity not only reduces charge transfer energy, but also is similar in size to exciton diffusion lengths (4–20 nm). Thus, when an exciton is generated within a PTB7 nanocrystallite, the process toward dissociating charges is greatly facilitated (inset of Figure 8). Whether or not this proposed morphology is inherent to PTB polymers or due to fluorine has yet to be determined. Thus, other fluorinated systems, especially their morphology, should be further investigated.
Similar improvements in morphology are observed in the benzothiadiazole and benzotriazole-based polymers, both of which were fabricated without the use of additives.21, 22 When compared with their nonfluorinated counterparts, the x-ray diffraction (XRD) data of PBnDT–DTffBT and PBnDT–FTAZ both show larger d-spacing values: 18.1 versus 17.7 Å for benzothiadiazole polymers and 18.7 versus 17.8 Å for benzotriazole polymers. It is likely that the repulsive nature of the fluorine atoms is keeping PCBM further away during electron-transfer reactions, possibly enhancing electron-hole charge-transfer complex separation and slowing down processes such as charge recombination. However, additional studies beyond XRD are needed to accurately elucidate the behavior between fluorinated polymers with PCBM.
This then begs the question: is there a certain fluorine concentration that leads to optimum interactions with PCBM? Jen and co-workers76 examined nonfluoro-, monofluoro-, and difluoro-substituted benzothiadiazole polymers PIDT–BT, PIDT–FBT, and PIDT–DFBT (Table 5). As expected, the HOMO energy levels lowered and Voc increased with increasing fluorine concentration on the benzothiadiazole acceptor moiety. However, other properties such as Jsc, FF (fill factor), and hole mobility were roughly similar for all three polymers. Given that this is just one specific series, it would be interesting to see similar studies conducted on other systems. Such studies would gauge the influence of fluorine concentration on how polymers pack with fullerenes and the effect on charge recombination (geminate and bimolecular) to give further insight on charge transfer processes with PCBM.
Table 5. Photovoltaic properties of PIDT-BT, PIDT-FBT, PIDT-DFBT.
3.2. Substitution on the Electron-Rich Donor Moiety
Not all fluorine substitutions appear to be beneficial. When Yu and co-workers77 fluorinated the BnDT donor moiety (Scheme 1), solar cells performed poorly compared with PTB7.77 Similar to the previous strategy, fluorinating the BnDT unit was intended to fine-tune the HOMO level of PTB polymers. The resulting HOMO levels of PTBF2 and PTBF3 were indeed lowered by –0.26 and –0.33 eV, respectively. However, transmission electron micrographs (TEM) of the polymer–PCBM films for PTBF2 and PTBF3 revealed noncontinuous networks with large phase domains on the order of 50–200 nm (Figure 9), encouraging charge recombination and leading to dramatic decreases in Voc, FF, and efficiency. In addition to the difficulty of synthesizing the fluorinated BnDT unit, PTBF2 and PTBF3 were observed to be unstable. The fluorines on BnDT pull electron density away from the TT moiety, concentrating it on the 4- and 6-positions of TT, making the polymer vulnerable to singlet oxygen attack.
3.3. Substituent Location
The improvement or decline in morphology of DONOR polymers is most likely related the location of the fluorine(s), more specifically which moiety is fluorinated. When the most electron-deficient unit is fluorinated (such as TT,11, 74 benzothiadiazole,21 or benzotriazole),22 the fluorines seem to keep PCBM at a distance creating phase domains (≈10–20 nm) that favor charge separation. It is unclear if this is a property inherent to these specific polymer systems because this favorable polymer–PCBM interaction is not observed when the electron-rich unit (BnDT) is fluorinated.77 From an electronic standpoint, this is in agreement with the weak donor-strong acceptor approach.78 The “weak donor” should be kept electron-rich and the “strong acceptor” should be as electron- deficient as possible.
In addition, a recent report by the Yu group suggests that electron-withdrawing groups should be placed such that the resulting local dipole moments do not cancel each other out based on their study of PTBF2 and PBB3.79 PTBF2 contains two opposing fluorines on the BnDT unit while PBB3 contains two adjacent TT units trans to another. In both cases, the internal dipole moment is greatly reduced according to calculations. Similar to PTB7, polymer PBB3 exhibits a good thin-film morphology, a high hole mobility, and even lower band gap (Table 6). Despite these favorable characteristics, PBB3 shows a comparatively low Jsc and thus efficiency of only 2.04%, suggesting that other factors need to be considered. Yu et al. propose that the minimized dipole moment in PTBF2 and PBB3 prevents the excited state from polarizing, leading to faster charge recombination and ultimately low power conversion efficiencies.
Table 6. Photovoltaic properties of PTB7, PTBF2, and PBB3.
The influence of fluorine on hole mobility, morphology, and other photovoltaic properties has yet to be quantified or correlated. Yu and co-workers65 suggest that there appears to be increased interaction between electron-rich aromatic rings and electron-deficient fluorinated aromatic rings. This is consistent with findings that fluorinated and nonfluorinated rings stack co-facially rather than in herringbone fashion as observed in traditional benzene rings.80, 81 Matsuo and co-workers82 have recently demonstrated that Ar—F··Ar—H and CH··F interactions help facilitate face-to-face stacking in FPPT compared with that in PPT (Scheme 2 and Figure 10), which leads to a hole mobility two orders of magnitude greater in FPPT.82 Although this study was done on small molecules for organic thin-film transistors, an analogous study in the context of DONOR polymers for solar cells would certainly be beneficial to further understand the interesting behavior of these fluorines. For example, would it be beneficial to have a 1:1 ratio of fluorinated to nonfluorinated rings? Would it be favorable for the donor and fluorinated acceptor moieties to be similarly shaped? More studies focused on the physical chemistry and device physics of carefully crafted systems are needed to elucidate fluorinated polymer–PCBM interactions and how morphology, hole mobility, local dipole moments, and charge recombination are affected.
4. Side Chains: Beyond the Solubility
One of the main advantages that organic solar cells can boast over their inorganic counterparts is that they can be solution processed, and therefore much cheaper to produce. Thus, side chains are a necessary component to designing conjugated polymers. Recent studies have discovered that the function of these side chains is for more than just solubilizing purposes. The nature of side chains employed often dictates the solid-state morphology in the active layer, which in turn, influences intermolecular interactions such as polymer–polymer and polymer–PCBM, as well as charge transport.48, 83 Inspecting the top polymers over 7% (Table 1) reveals no clear pattern of the best combination of side chains and where on the backbone they should be anchored. The optimum combination of position and size is likely to be polymer specific and sometimes can only be determined after synthesizing an exhaustive library. Nevertheless, this section will attempt to survey key guidelines that have emerged as generally applicable, and shine a spotlight on less commonly employed chains by examining the following types: nonaromatic, aromatic, and end-group functionalized.
4.1. Nonaromatic Side Chains
The vast majority of DONOR polymers utilize simple alkyl or alkoxy side chains, and deciding where to position them on the polymer can profoundly affect performance. The PBDT–DTBT series demonstrates that the optimum location for side chains should cause the least steric disturbance to the planarity of the polymer backbone.84, 85 In this series, PBDT–4DTBT, which is alkylated at the four-position of the thienyl groups, exhibited the highest efficiency in its BHJ solar cells (Table 7). Similar to the control polymer (nonalkylated PBDT–DTBT), PBDT–4DTBT maintains the most planar backbone as evidenced by its small calculated dihedral angles and low band gap. But unlike the control polymer, PBDT–4DTBT's solubilizing chains allow it to achieve a higher molecular weight and efficiency. Since many DONOR polymers contain thienyl groups, the design concepts established in this work can easily be applied to those systems as well as others. This study highlights the importance of strategically placing solubilizing chains such that there is no excessive twisting in the backbone and polymers can attain high molecular weight.
Table 7. Power conversion efficiencies, calculated dihedral angles, and polymerization results for PBDT–DTBT polymers. Reproduced with permission. Copyright 2010, American Chemical Society.
Upon deciding where to place the side chains, the next decision is what length (long or short) and shape (linear or branched) they should be, which can greatly impact properties such as Jsc and Voc. You and co-authors86 studied six polymers with an identical backbone (PNDT–DTBT) but with varying linear and branched side chains on both the NDT and DTBT units (Table 8).86 Because of the identical backbone, the different side chain combinations represent the difference in π–π stacking between the aromatic cores. In general, a closer π-stacking distance reduces the energy barrier for intermolecular charge hopping while also minimizing charge trapping sites.87 This systematic study on PNDT–DTBT polymers demonstrates that long and branched side chains weaken the intermolecular polymer interactions but also enhance Voc (polymer C10,6-C6,2). On the other hand, short and straight side chains encourage polymer packing, increasing the Jsc at the expense of Voc (polymer C8-C6,2). In order to mediate these opposing trends, it was found that short and branched side chains (polymer C6,2-C6,2) are the best compromise for attaining reasonably high Voc and Jsc, leading to the optimum efficiency of 3.36% in this series.86 A similar side chain study by Fréchet and co-workers88 found that longer linear side chains can be used in place of branched chains for more soluble cores such as the furan-diketopyrrolopyrrole system.
Table 8. Photovoltaic properties of PNDT–DTBT polymers. Reproduced with permission. Copyright 2010, American Chemical Society.
Yu and co-workers74 also found that linear versus branched chains affected polymer packing in the PTB polymers. As previously mentioned, PTB polymers intermolecularly stack in a face-on orientation. This favorable packing can be enhanced depending on whether or not the side chains are branched. GIWAXS results revealed that the BnDT unit is mostly responsible for controlling intermolecular π–π stacking interactions as it is composed of three fused aromatic units. Therefore, branched side chains on this unit increase the π–π stacking distance, decreasing FF and efficiency. For instance, the structures of PTB1 and PTB5 differ greatly by the chains on the BnDT unit (Figure 11). PTB1 containing a linear side chain exhibited a 3.65 Å π–π distance and 5.6% efficiency, whereas PTB5 containing a branched chain exhibited a larger 3.89 Å π–π distance and lower efficiency of 4.1%. In contrast, the side chain type on the TT unit does not appear to influence intermolecular π–π stacking, but most likely does so with PCBM interactions. For example, PTB1 and PTB2 contain the same chains on BnDT but linear or branched side chains, respectively, on the TT moiety, yet both exhibit the same 3.65 Å π–π spacing. In a similar side chain study on benzodithiophene and diketopyrrolopyrrole-based (BnDT–DPP) copolymers,89 Li et al. proposed that the electron-rich BnDT should contain a linear side chain to possibly increase its contact with electron-poor PCBM and enhance charge transfer. Meanwhile the electron- deficient moiety DPP should contain bulky branched side chains to most likely repel PCBM and therefore prevent charge recombination (Figure 12). Thus, polymer O-HD was the front-runner in terms of photovoltaic performance (Table 9).
Table 9. Photovoltaic properties of BnDT–DPP polymers. Reproduced with permission. Copyright 2011, American Chemical Society.
Despite these insightful studies on the type of side chains that should be used and where they should be anchored on the backbone of conjugated polymers, finding the optimum combination is still very much polymer specific and likely still an empirical process. For example, Fréchet and co-workers48 investigated a series of copolymers (PBnDT–TPD) based on the BnDT and N-alkylthieno[3,4-c]pyrrole-4,6-dione (TPD) (Scheme 3). According to grazing incidence X-ray scattering (GIXS) studies, PBnDT–TPD polymers may also pack face-on toward the substrate. However, unlike the TT in the previously mentioned PTB series, chain length on TPD moiety did in fact influence π–π stacking in the PBnDT–TPD series. The ethylhexyloxy chain on the BnDT was kept constant where R was varied on the TPD moiety. PBnDT–TPD1, which contained a short and branched ethylhexyl chain showed a larger π-stacking distance of 3.8 Å, whereas PBnDT–TPD2 and PBnDT–TPD3, which contained dimethyloctyl and octyl chains, respectively, showed a smaller d-spacing of 3.6 Å and lower efficiencies in their BHJ devices.
4.2. Aromatic Side Chains
Although much effort has gone into determining position, length, and branching of these solubilizing alkyl chains, the research field developing nonalkyl solubilizing chains, still remains under-explored. Aromatic side chains are particularly attractive because they can extend the conjugation of the polymer and therefore possibly promote hole mobility. Huo et al.14, 90 reported a series of PBDTTT polymers which compare alkylthienyl side chains against alkoxy chains (Table 10).14, 90 Both of the alkylthienyl-substituted polymers exhibited smaller band gaps, larger Jsc values, and higher efficiencies. The higher Jsc values were attributed to the higher hole mobilities of these polymers. These results indicate that although aromatic units as side chains may cause steric hindrance, this steric bulk can be advantageous if it extends conjugation and does not cause excessive repulsion between the polymer and PCBM.
Table 10. Photovoltaic properties of the PBDTTT polymer series. Reproduced with permission.
4.3. End-Group Functionalized Side Chains
As charge separation occurs at the DONOR-ACCEPTOR interface, the physical interaction between the polymer and PCBM has a significant impact on device performance. Kim et al.91 have demonstrated that there is a correlation between end-group modification and the morphology of the active layer.91 In their study, P3HT polymers end-capped with –Br, –OH, –CH2CH3, and –CF2CF2CF3 were examined (Table 11). Of the four, P3HT–CH3 and P3HT–CF3 showed the highest efficiencies in BHJ devices and closest surface energies to PCBM. These well-matched surface energies resulted in a favorable balance between miscibility and phase separation between the polymer and PCBM, which ultimately led to better charge transport across the DONOR-ACCEPTOR interfaces. This suggests precedence that fluorinated polymers are morphologically compatible with PCBM.
Table 11. Photovoltaic parameters of end-functional-group modified P3HT. Adapted with permission.
So far there are a limited number of side chain end-modified polymer examples. However, the following squarine dyes and diketopyrrolopyrrole molecules (Scheme 4) have shown promising results. Bagnis et al.92 compared two new alkyl- and alkyenyl-functionalized squarine dyes and found that both molecules had the same energy levels, band gap, and morphology.92 However, the noncovalent alkenyl–phenyl interactions in Sqr. 2 resulted in a smaller stacking distance between the cores. This then translated into a higher mobility and higher Jsc. In a similar fashion, Bao and co-workers93 reported that the siloxane-terminated solubilizing chains of diketopyrrolopyrrole-based molecule (PII2T-Si) for thin-film transistors decreased the π–π stacking between adjacent molecules and boosted the hole mobility. In light of these results, employing alkyenyl, siloxane, or end groups with similar effects are certainly synthetically feasible and could likely improve hole mobility.
Although deciding the appropriate side chains for specific polymers is an empirical process that will no doubt require the creation of libraries to discover the best combination, certain key guidelines have indeed emerged. The first is that the location of the side chains should be such that they cause the least amount of twisting along the backbone. The second is that in general, short and branched side chains seem to concurrently provide the optimal Jsc and Voc. And finally, both aromatic and end-functionalized side chains have shown the potential to enhance hole mobility through strengthening intermolecular polymer–polymer interactions. Thus, alkenyl-, siloxane-terminated, and alkylthienyl side chains should be further explored as well as many others. There are other important issues regarding side chains, which are beyond the scope of this discussion but nevertheless merit further consideration, including but not limited to side chain density94 and fluorinated side chains for use as interfacial additive layers.95
Pushing the power conversion efficiency past 10% is certainly within the OPV community's reach, however doing so will require synergistic efforts from synthetic chemists to physicists to engineers to tackle current hurdles. The issues mentioned in this article are not meant to be an exhaustive laundry list, but rather a few key topics to inspire further studies. An emerging theme is the need for a better understanding of correlating structure–property relationships. Although much emphasis is often placed on lowering the HOMO level to increase Voc, focusing on developing LUMO-reducing moieties would recover wasted energy during electron transfer to PCBM while enhancing Voc and Jsc as well. Studies elucidating how PCBM interacts with the polymer are also especially vital. Various studies have implied that electron-poor PCBM should be kept near the electron-rich moiety of the polymer to enhance charge dissociation, whereas PCBM should be kept away from the electron-poor moiety to prevent charge recombination. These proposed models have yet to be proven or disproven; however, studies have demonstrated that both fluorine and side chain substituents can tune these interactions. If such structure–property relationships can be determined, this will greatly aid toward the movement of reaching 10% efficiency and beyond.
We thank the National Science Foundation for supporting this work through a CAREER Award (DMR-0954280) and Grant CHE-1058626. W.Y. is a Camille-Dreyfus Teacher-Scholar.
Rycel Uy earned her B.S. in Chemistry from the University of Nevada, Las Vegas in 2008. She is currently a Ph.D. candidate in Professor Wei You's group at the University of North Carolina at Chapel Hill, where she works on developing new polymer materials, particularly thienothiazole-based ones, for use in bulk heterojunction solar cells.
Sam Price earned his B.S. in Chemical Engineering from North Carolina State University in 2006, and received his Ph.D. in Chemistry in Prof. You's group in 2011 studying conjugated polymers. He is currently a postdoctoral researcher for the Army Research Lab at Aberdeen Proving Ground. His research interests focus on functional materials for energy and electronics applications.
Wei You was born in a small village outside of Chuzhou in Anhui Province of China, and grew up in Hefei, the provincial capital of Anhui. After receiving a B.S. degree in Polymer Chemistry from University of Science and Technology of China in 1999, he attended the graduate program of chemistry at the University of Chicago, where he obtained his Ph.D. in 2004 under the guidance of Professor Luping Yu. He then moved west and finished his postdoctoral training at Stanford University in 2006 with Professor Zhenan Bao. In July 2006, he joined the University of North Carolina at Chapel Hill as an Assistant Professor in Chemistry. Professor You's research interests focus on the development of novel multifunctional materials for a variety of applications, including organic solar cells, molecular electronics, and spintronics.