The past five years has witnessed an explosive growth in the area of organic photovoltaic cells (OPVs). A web of science search using keywords “OPVs or organic solar cells” results in about 1750 peer-reviewed publications in 2011. The reported power conversion efficiencies (PCE) in OPVs have increased from about ∼5% in 2005 to ∼9.8% in 2011.1 This increase in PCE can be largely attributed to the development of new conjugated polymers with a low band gap. The continued focus of research in OPVs is on achieving efficiencies greater than 10%.2 This target seems much closer now than five years ago for laboratory-scale cells. However, the pace toward this target in commercial cells has been slow. In this review, we outline some of the challenges in achieving higher PCEs. In particular, we will focus on our current understanding of the morphology of the active layer and the approaches toward obtain nanoscale phase segregated active layer morphologies.
Typically, the active layer of an OPV consists of conjugated polymer that acts as the electron donor and a fullerene-based molecule (PC60BM or PC70BM) that acts as the electron acceptor.3–5 The photons are absorbed by the conjugated polymer, which creates excited electron-hole pairs. These electron-hole pairs, excitons, either recombine to emit light or heat or, in a more useful process, dissociate and move to their respective electrodes to result in a photovoltaic current. To achieve higher efficiency in solar cells, the absorption spectrum of the active material(s) should overlap with the solar emission spectrum (AM 1.5) and the following requirements have to be met: (a) hole and electron transporters with optimal frontier energy levels, should be assembled into structures such that heterojunctions should exist at lengths scales of a few nanometers for efficient charge separation, (b) continuous phases of charge-carrier transporters should exist for efficient charge transport, and (c) the active medium (i.e., the thickness of the OPV device) should have a characteristic length scale of a few hundred nanometers to match the typical photon capture distance. The first challenge therefore is to develop a donor material that has a broad absorbance in the solar spectrum with appropriate frontier orbital energy levels for optimal charge transfer. The archetypical donor polymer regioregular poly(3-hexythiophene) (P3HT) has a band gap of 2.1 eV with the highest occupied molecular orbital (HOMO) level at 5.1 eV and lowest unoccupied molecular orbital (LUMO) level at 3.2 eV with a reported PCE about 3–5%.6–8 It has been argued that an ideal donor polymer (with PCBM as the acceptor) should have a band gap of 1.5 eV with the HOMO level around 5.4 eV and LUMO level around 3.9 eV.7 In the past five years, a plethora of thiophene-based conjugated polymers have been synthesized with varied band gaps and energy levels (see Figs. 1 and 2).9, 10 Few of these polymers have also been reported to have high PCEs.10–17 Whereas OPV devices with higher efficiency have low band gap polymers as active material, there is no guarantee that using low band gap polymers will result in high efficiencies (Fig. 3). The second challenge therefore is to assemble hole transporters and electron transporters into morphologies for efficient charge separation and transport. This topic is the focus of this review.
MORPHOLOGIES IN OPV ACTIVE LAYERS
The common approach to creating active layer morphologies for organic solar cells is to cast a film from a solution containing a conjugated polymer with an electron transporter (organic or inorganic) and then thermally or solvent anneal the thin films over a specific period of time.18 The target morphology – termed as bulk heterojunction (BHJ) – has bicontinuous nanoscale domains of the charge conductors throughout the active material for charge separation and charge transport.4, 19, 20 The P3HT–PCBM blend is one of a few systems that is known to form this morphology (see Fig. 4) and therefore has been subjected to numerous investigations.6, 8, 21–51 These studies have provided an in-depth understanding about the evolution of this morphology. For example, HR TEM images obtained by Russell and coworkers,49 clearly show that P3HT and PCBM are homogenously distributed before the annealing process [Fig. 5(a)]. The segmental interaction parameter (χ) between P3HT and PCBM calculated from the depression in melting point was found to be −0.162.49, 50 This is consistent with the fact that P3HT is electron rich and PCBM is electron poor. Therefore, the interaction between P3HT and PCBM is enthalpically favorable. HR TEM images after annealing, shows the appearance of regions rich in P3HT and in PCBM [Fig. 5(b)]. During annealing, X-ray diffraction patterns show enhanced P3HT crystallinity and PCBM gets concentrated in areas where P3HT is amorphous. The diffusion of PCBM is extremely rapid during the annealing process with an estimated diffusion coefficient of 10−10 cm2/s.50 The morphological features that develop during annealing are kinetically trapped when the annealing is stopped. Prolonged annealing, leads to macrophase segregation of P3HT and PCBM domains, which degrades the device performance. It is now well understood that there are three factors that concurrently facilitate the formation of a BHJ in the P3HT–PCBM system.3, 49, 50 These are (1) the propensity of P3HT to rapidly crystallize in the presence of PCBM, (2) the large diffusion coefficient of PCBM in amorphous P3HT and (3) the slow kinetics of crystallization of PCBM in the presence of P3HT. It is noteworthy that the HR TEM images reported by Russell and coworkers49 also show that the interface between PEDOT/PSS and P3HT becomes stronger after the annealing process.
Much like annealed P3HT/PBCM films, annealed P3HT films without PCBM also contain crystalline and amorphous domains. Therefore, if our understanding of the BHJ is correct then PCBM should diffuse into P3HT films. It has been shown that sequential deposition of P3HT and PCBM (bilayered morphology), upon annealing indeed leads to the same BHJ morphology as the one obtained from a solution containing P3HT and PCBM (Figs. 6 and 7).48, 52
Although P3HT/PCBM system has been well-studied and recent studies have provided an in-depth understanding of the BHJ morphology, it has not resulted in a general protocol to obtain BHJ morphologies. Very few conjugated polymers have the propensity to crystallize in the presence of an electron acceptor since there is an attractive interaction between the conjugated polymer and the electron conductors.53–55 Very few electron conductors have the diffusion and crystallization characteristics of PCBM. The fundamental limitation of the annealing process to obtain BHJ morphology is that the formation of key features such as domain sizes and interfaces that make the BHJ morphology important for OPVs intricately depends on interdependent factors such as the packing of the conjugated polymer in the presence of an electron acceptor, which cannot be directly varied or controlled. Thus, one cannot exercise control and tune the domain sizes or the interface for optimal device performance. Therefore, there is a need for a method to reliably and reproducibly organize electron- and hole-conducting semiconductors into nanoscale BHJ type structures that provide efficient charge separation and charge transport. In the ensuing sections of this review, we focus on some of the efforts toward a general protocol for controlling the morphology.
USE OF ADDITIVES TO CONTROL ACTIVE LAYER MORPHOLOGIES
In 2006, Bazan, Moses, and coworkers56 reported that in P3HT/PCBM system, addition of n-octylthiol dramatically enhanced the structural order in the P3HT domain. As a result, the hole mobility increased by two orders of magnitude. Based on this observation, in 2007, Bazan, Heeger, and coworkers reported that in the PCPDTBT/C71-PCBM system, addition of 1,8-octanedithiol increased the PCE from 2.8 to 5.5% (Fig. 8).54 The assumption was that much like in P3HT, 1,8-octanedithiol may help in increasing the structural order in PCPDTBT. However, in contrast to the P3HT/PCBM systems, X-ray diffraction indicated that there was no increase in the structural order. FET mobility measurements indicated that there was no increase in the hole mobility of PCPDTBT. It has also been observed that the additive should be completely removed under vacuum as they can act as hole traps thereby reducing the hole mobility. The additive, however, was reported to enhance the electron mobility.53 The observation that 1,8-octanedithiol can enhance the PCE led to a flurry of investigations into other possible additives, which identified 1,8-diiodooctane (DIO) as an effective additive to enhance PCEs.55 In these protocols, typically, the conjugated polymer and PCBM are dissolved in chlorobenzene or 1,2-dichlorobenzene containing 2.5% (vol %) of the additive and this solution is then spin cast over desired substrate.
How does the additive enhance the PCEs? In 2008, Bazan, Heeger, and coworkers concluded that the additive leads to the formation of three phases during the film drying process: (a) additive-PCBM phase (b) polymer aggregate phase and (c) polymer-PCBM phase. As the additive has a higher boiling point than the host solvent and the additive is known to preferentially solubilize PCBM, they concluded that during the drying process, PCBM stays longer in solution than the polymer.53 This may allow the PCBM to reorganize and result in a better nanophase segregated structures with appropriate domain sizes. For an additive to be effective, they postulated that the additive should preferentially solubilize PCBM over the conjugate polymer and should have a boiling point higher than 1,2-dichlorobenzene.
More recently, Marks and coworkers57 studied the effect of DIO in the PTB7/PC71BM system using solution phase small angle X-ray scattering. Their observations can be summarized as follows: (a) without DIO, large aggregates of PC71BM are observed with mean radius of 11 Å (b) addition of DIO has little effect on the size of polymer aggregates (mean radius of 34 Å without DIO and mean radius of 36 Å with DIO) and (c) addition of DIO leads to the reduction of the PC71BM aggregate size from ∼11.5 to ∼5 Å (Fig. 9); DIO also decreases the number of PC71BM aggregates. From these observations, they concluded that DIO selectively solubilizes PC71BM thus facilitating the incorporation of PC71BM into the PTB7 aggregates (Fig. 10). The use of additives provides a straightforward way to obtain active layer morphologies through spin casting without need for thermal or solvent annealing. There seems to be a general agreement that additives solubilize PCBM, suppresses the formation of large aggregates thus providing control on the domain size of PCBM. However, the effect of additives on the size of the polymer aggregate is unclear. There are reports that indicate that the additives facilitate the aggregation of the polymers and there are others that indicate that there is little effect on the aggregation of the polymers.55 The logical conclusion here is that the impact of additives on the polymer aggregation is case specific and variable. Thus, there is a need for an effective method to control the size of the polymer domains.
USE OF NANOWIRES
A popular approach to control the polymer domain size is to preaggregate the polymer into fibers or wires. It has been known that aggregation of the conjugated polymer can be induced by adding a poor solvent for the polymer. In 2007, Guillerez and coworkers58 reported that using preformed P3HT fibers and PCBM, PCE of 3.6% can be obtained without annealing. In comparison, films made from amorphous P3HT and PCBM showed a PCE of 0.65%. The P3HT fibers were obtained by heating a 1 wt % P3HT solution in p-xylene, cooling it to room temperature at the rate of 20 °C/h and allowing this solution to stand at room temperature for an extended period of time (Fig. 11). The fibers were separated from low molecular weight fractions by centrifugation and by filtration. The typical dimension of the fibers were 0.5–5 μm in length, 30–50 nm in width and 5–15 nm in thickness. The authors then studied the effect of mixing the P3HT fibers with amorphous P3HT in various proportions and measured the PCE (Fig. 12). They found that the maximum PCE was obtained in films containing 75% crystalline P3HT fibers and 25% amorphous P3HT.58 This observation that 25% of amorphous P3HT is required for maximum PCE is consistent with our current understanding of the BHJ morphology in which the PCBM concentrates in amorphous P3HT domains. Interestingly, the device fabricated with 97% crystalline P3HT fibers showed a respectable PCE of 2.1%. This observation indicates that the PCBM may also be accommodated at the interfaces of the crystalline domains or the P3HT fibers may have regions with poor-crystalline or amorphous P3HT. This work laid the groundwork for controlling the conjugated polymer domain and obtained BHJ morphology without thermal annealing thereby reducing the number of steps involved in the fabrication process.
Controlling the nanofiber dimensions offers control over the conjugated polymer domain sizes. But for high efficiencies, crystallinity of the conjugated polymer within the nanofiber is also need to be high. For example, long P3HT fibers can be made from anisole and devices fabricated from these fibers and PCBM showed photovoltaic performance that was inferior to fibers obtained from p-xylene.59 However, the authors found that addition of small amounts of chlorobenzene vastly improves the crystallinity of the fibers and thus the overall photovoltaic metrics. In 2011, Cho and coworkers60 reported that P3HT nanowires obtained from dichloromethane showed increased crystallinity upon ageing. They studied the effect of ageing for 12–72 h in 12 h increments. They found that maximum PCE of 3.2% was obtained after 60 h of ageing. The open circuit voltage and fill factor did not change over ageing but the Jsc showed an increase with respect to ageing time. The authors measured the hole and electron mobility in films obtained from these samples. They found that, upon ageing, there was an increase in the hole mobility but the electron mobility showed only marginal changes. More importantly, the ratio of the electron and hole mobility decreased with respect to increased ageing time. The maximum PCE was obtained when the mobility ratio (μe/μh) was 1.4 [see ageing time 60 h in Fig. 13 (right)]. This work clearly illustrates the hole mobility can be increased by enhancing the crystallinity of nanofibers and the need to match electron and hole mobilities in active layer thin films for high PCEs.
In 2008, Jenekhe and coworkers61 reported PCE of 3% in films obtained from poly(3-butylthiophene) (P3BT) fibers/nanowires and PC71BM. The nanowires were about 8–10 nm in width and about 5–10 μm in length. The PCE of devices fabricated from thin films obtained by spin casting a solution containing P3BT and PC61BM followed by thermal annealing showed a PCE of 1%. The films obtained with P3BT fibers were thermally annealed at 110 °C for 5 min. In comparison, the PCE for P3HT nanowire-fullerene systems in unannealed samples was 3.6%. In 2010 Jenekhe and coworkers62 reported the impact of various processing conditions on the PCEs in the P3BT nanowires/PC71BM system. They found that the thermal annealing of the films at 175 °C for 10 min increases the efficiency from ∼1 to ∼3%. They also found that after the films were cast, the drying time before annealing played an important role in the PCEs. When the P3BT nanowires/PC71BM films were dried for 3 min before annealing (condition A), AFM and TEM showed little evidence of nanowire presence (Fig. 14). The devices made from these films after annealing showed PCE of 3.10%. Conversely, when the P3BT nanowires/PC71BM films left for 100 min before annealing (condition D), then AFM and TEM showed the presence of large P3BT nanowires (Fig. 14). The devices made from these films after annealing showed PCE of 3%. In comparison, devices from unannealed films showed PCE of 1.18%. Interestingly, the space-charge limited current (SCLC) hole mobility in films dried for 3 min was found to be higher than the films with a drying time of 100 min. Recall that in P3HT/PCBM system, the ageing increased SCLC mobility of the fibers. The observations in the P3HT and P3BT indicate that the formation of nanowires may provide a pathway to control the size of the conjugate polymer aggregates. But controlling the crystallinity of the conjugated polymer in nanofibers and optimizing the processing conditions are necessary to achieve high photovoltaic efficiencies. The next step would be to combine the nanowire approach with the use of additives, which may provide a path to control the domain sizes and assembly of both the conjugated polymer and PCBM.
USE OF NANOPARTICLES
The prior section dealt with the aggregation of conjugated polymers to nanowires or nanofibers. In 2002, Landfester, Scherf, Neher and coworkers64 reported the use of miniemulsion process to create conjugated polymer semiconductor nanospheres. In this procedure, the conjugated polymer is dissolved in a good organic solvent and is added to water containing a surfactant. Ultrasonication of this emulsion followed by the evaporation of the solvents results in dispersion of conjugated polymer nanoparticles in water. As this discovery, several conjugated polymers have been formulated as nanoparticles using miniemulsion, reprecipitation, or other methods. In 2003, Landfester, Scherf, Neher, and coworkers65, 66 reported that devices fabricated by blending SDS-stabilized nanoparticles derived from PFB, a hole-transporting polymer and F8BT, an electron-accepting polymer showed an maximum external quantum efficiency of 1.5%. They also showed that the efficiency can be increased to 4% if in the miniemulsion process chloroform was used instead of xylene to dissolve the polymers.65, 66 They also showed that efficient light emitting diodes can be fabricated from SDS-stabilized nanoparticles and that the influence of SDS is minor on device performance.67 These reports indicate that the surfactants may not adversely affect charge transport in stabilized nanoparticles. Although the reported efficiency is lower than cells fabricated by other methods, the use nanoparticles may lead to a more general approach to control the domain size and morphology. It may also open up new avenues in terms of morphology.
A straightforward way to control the size of the domains is by controlling the size of the nanoparticle. Indeed varying parameters such as surfactant concentration in the miniemulsion process or the ageing time in the reprecipitation method can result in particles with different mean sizes.68, 69 For example, in the preparation of alkyl-PTA and alkoxy-PTA nanoparticles by reprecipation method, Lai and coworkers70 observed that nanoparticle sizes grow over time (Fig. 15). If the nanoparticles derived from hole-conducting conjugated polymers are similar in size to nanoparticles derived from electron-accepting conjugated polymers, in principle, then one can print layer by layer by blending the nanoparticples appropriately to result in graded heterojunctions (Fig. 16).
What if the nanoparticles are not of the same size? Would blend nanoparticles result in morphologies that will be useful for OPVs? To answer this question, we turn to the rationalization for the formation of ionic mineral structures. In 1929, Pauling formally enunciated the radius ratio (Rcation/Ranion) rules for predicting the coordination number and thus the structures of ionic crystals.71 These rules, though approximate, are powerful in predicting the possible structures that can result from a anion–cation pairs by treating the ions as hard spheres. Akin to the formation of binary ionic crystals, two types of nanoparticles can also self-assemble into ordered “superlattices” at a much larger scale.72–74 The assembly of the nanoparticles also seem to be dictated by the ratio of the radii of the particles (γ = Rsmall/Rlarge). More interestingly, the self-assembly of nanoparticle mixtures occur even if the nanoparticles do not have electrostatic attraction.74, 75 The primary mechanism that drives assembly of spherical particles into crystalline arrays (or superlattices) is maximization of entropy by adopting the most efficient packing. The classic example of entropy-driven crystallization is the formation of face-centered cubic (FCC) arrays of hard spheres (having billiard-ball-like interactions). The FCC lattice has the largest close-packed particle volume fraction, ϕ, of any ordered lattice containing spheres of the same size: ϕ = 0.74. Hence, this lattice provides more free volume to each particle than does a liquid (which is comparatively inefficiently packed and has a maximum ϕ ∼ 0.64).76 Computer simulations75, theory,77–79 and experiments with micron-size colloidal spheres show that the same principle applies in mixtures of two different sizes, in which AB2 and AB13 (AlB2 and NaZn13) superlattices were found in systems with size ratios γ approximately 0.58 and 0.62.80–82 Recent investigations of binary nanoparticle mixtures with 0.57 < γ < 0.81 have shown these and other structures and they are summarized in Figure 17.72, 74, 76, 83
We turn our attention to structures that are relevant for solar cells. In Figure 17, we show the commonly observed structures in inorganic nanoparticle assemblies.72 For example, the mineral structure AlB2 consists of alternating layers of Al and B atoms (Fig. 18 left). This structure-type has been obtained by self-assembly of 6.7 nm PbS and 3 nm Pd nanoparticles.72 Now consider the possibility of replacing the PbS and Pd nanoparticles in the AlB2 structure with electron-transporting and hole-transporting nanoparticles. The resultant assembly will have a lamellar arrangement of semiconductors, that is, alternating layers of hole and electron transporters – the most widely targeted structure of PV cells (Fig. 18, right). Similarly, CaCu5 and MgZn2 structure types (Fig. 17) have continuous phases of each species of nanoparticle, with an interface between the two phases. By varying the radius ratio of the particles, one can potential tune the active layer morphologies of photovoltaic devices. The diversity of binary lattices creates multiple opportunities to optimize heterojunction structures and therefore photovoltaic efficiency.
The use of nanoparticles has several advantages over existing methods for directing the active layer morphology. The use of nanoparticles combines the natural propensity of the electron–donor moieties to mix with electron–acceptor moieties with packing propensities of spheres to obtain structures relevant for PV devices. Thus, this approach may allow the formation of stable heterojunction structures in a single step through self-assembly of spheres. Second, one can tune the packing geometry of the semiconductor assemblies by varying the radius ratio of the spheres. Third, one can tune the characteristic length scale by varying the radii of the spheres but keeping the radius ratio constant. Fourth, this strategy to direct the packing of semiconductors to specific morphologies may enable us to answer fundamental questions in the area of photovoltaic devices, including the optimal structure for an efficient PV device. Finally, the knowledge can be easily extended to nanoparticles of small molecules or electron conducting polymers.
There are several barriers that need to be removed to realize full potential of this approach thus providing fantastic research opportunities. Typically, the formation of crystalline structures requires narrow size dispersion in particle size. The synthetic protocols for inorganic nanoparticle do result in narrow size distribution whereas protocols for organic semiconductor nanoparticles provide a broader size distribution. The challenge is to optimize these methods to result in narrow size distribution of particles. The second challenge is to control the packing of the conjugated polymer within the nanoparticle and the impact of this packing on the photophysical and electronic properties. Recall that the photophysical properties of P3HT nanofibers from anisole are different fromP3HT nanofiber from p-xylene.59 Similarly, the F8BT/PFB nanoparticles obtained from CHCl3 gave different EQE values than the particles obtained from xylene.65, 66 Only recently have researchers started to probe the impact of the packing of the polymer within the nanoparticle on the photophysical properties.76, 84 The third challenge is the presence of surfactant molecules that stabilize the nanoparticles. The charged surfactants can help the self-assembly process through ionic interactions. The conventional wisdom is that the surfactant on the nanoparticle may impede charge transport. Thus far, one report indicates that the effect of surfactant on charge transport in minimal. If our conventional wisdom turns out to be correct that the surfactants sufficiently impacts charge transport, then synthetic protocols needs to be tuned to provide stable nanoparticle dispersions that can facilitate charge-transport in condensed phase assembly. The final challenge will be to obtain assemblies from a blend of nanoparticles that allows charge transfer and charge transport. Initial reports by Landfester, Scherf, and Neher combined with recent observations in inorganic nanoparticle assemblies provide us the necessary optimism in this burgeoning area.
In this review, we have outlined methods to control the active layer morphology in OPVs. The traditional method of spin cast followed by thermal annealing has provided some initial breakthroughs in OPV efficiencies. Recent studies have firmly established the evolution of the active layer morphology in this process. The use of additives and the use of nanowires have allowed us to build on the gains obtained by the thermal annealing process. Compared to these processes, the area of nanoparticle assembly is still in its infancy. Yet, it has the potential to be a disruptive approach to active layer morphology.
The authors thank Prof. Anthony Dinsmore and Prof. Michael D. Barnes for their valuable discussions on the subject of nanoparticles and their assembly. They thank PHaSE Energy Frontier Research Center supported by the US Department of Energy, Office of Basic Energy Sciences, through grant DE-SC0001087 (DMR 0820506) for financial support of this work.
G. Nagarjuna, Graduate Student. He did his Masters in chemistry from University of Hyderabad. Currently he is pursuing Ph.D under the guidance of Prof. D. Venkataraman at University of Massachusetts-Amherst. The focus of his research is synthesis and assembly of semiconducting conjugated nanoparticles and synthesis of novel n-type conjugated polymers.
D. Venkataraman is an associate professor of chemistry at the University of Massachusetts Amherst. His current research focuses on the synthesis and directed assembly of conjugated molecules and polymers for efficient charge transport. For further information, see http://thedvgroup.com