Recent advances in high-efficiency organic solar cells fabricated by eco-compatible solvents at relatively large-area scale

As one of the most promising technologies for harvesting clean and renewable solar energy, bulk heterojunction (BHJ) organic solar cells (OSCs) have aroused considerable attention over the past several decades because of their attractive merits of light weight, high flexibility, and suitability for large-scale solution-processing manufacturing.^1–3 Recently, the power conversion efficiencies (PCEs) for OSCs have been boosted to over 18% because of significant progress in the development of new electron-donor and electron-acceptor materials, showing great application prospects.^4–8 However, it should be noted that top-performing devices usually have very small areas (∼0.04 cm²) and are processed from highly toxic halogenated solvents, which becomes one of the main limiting factors on the road to realization of commercial applications for OSCs.^1,9–13 It has been proved that highly efficient materials and most widely used device processing technologies for lab-scale devices are incompatible with industrial-scale large area OSC modules.¹⁴ In terms of device preparation, spin-coating is the most popular film-forming method in the laboratory for its convenience, low price, and easy availability. However, due to its extremely low material utilization efficiency, it is not suitable for large-scale production. In addition, the present materials with excellent photovoltaic performance are optimized for obtaining the best blend morphology via spin-coating procedures. When using large-area technologies, the resulting devices always exhibit a large decrease in efficiency because of relatively bad blend morphologies for the active layers. Therefore, developing new materials that can be used to fabricate large-area and high-efficiency devices by coating or printing technologies is crucial for realizing practical applications of OSCs (Fig. 1). Furthermore, in addition to developing suitable film-forming techniques for building high-efficiency large-area OSCs, it is necessary to make sure that eco-compatible solvents are used in the manufacturing processes, which is very important for mass production.^15,16 If highly toxic solvents are used in the fabrication process, the risk of polluting the environment will increase, and thus, the use of such solvents will be forbidden by many countries. Unfortunately, in order to obtain good solubility for light-harvesting layer materials and favorable film morphologies, halogenated and/or aromatic processing solvents such as chlorobenzene (CB), chloroform (CF), and 1,8-diiodooctane (DIO, which usually works as an additive to modulate the film morphology) are widely used in the fabrication of OSCs with cutting-edge PCEs. Therefore, removing toxic solvents from the preparation process for OSCs is of vital importance for practical application in the near future.

In this Perspective, we begin with a brief introduction to OSCs, including typical large-area processing technologies and commonly used processing solvents. Then, we discuss the recent advances for highly efficient light-harvesting layer materials for large-area technologies using eco-compatible solvents, where new emerging non-fullerene acceptor (NFA) materials are highlighted. Finally, we summarize the critical issues that should be solved to meet the requirements for the commercialization of OSCs and provide an understanding for guiding the design of new materials for practical applications.

The solution processability of organic materials is one of the most attractive advantages for OSCs, which makes OSCs suitable for large-area production via scalable printing technologies that promise low cost.¹⁷ As shown in Fig. 2(a), a typical single-junction OSC device usually consists of two electrodes, interface layers, and a light-harvesting layer. The fabrication of the electrodes can follow the proven technologies used in OLEDs using vacuum evaporation methods. In addition, if a conductive glue can be successfully used to fabricate electrodes via large-area technologies, the cost of OSCs will significantly decrease. In addition to the functions of charge collection and transport, one of the electrodes should be transparent to ensure efficient utilization of solar photons by the active layer. The other electrode is usually aluminum or silver, which can reflect unabsorbed light to realize secondary utilization. In general, the two electrodes are modified with interface layers, namely, anode and cathode interfacial layers, to optimize the electronic and electrical properties among the interfaces of the light-harvesting active layer and the charge-collecting electrodes. It should be noted that the widely used interfacial layer materials such as PEDOT:PSS, PFN-Br, PDINO, and sol–gel ZnO are usually soluble in eco-compatible solvents such as water or methanol.^18,19 Moreover, slot-die coated interlayers have been achieved by Madsen et al.²⁰ However, large-area film processing methods still need to be further developed, which can form a desirable surface morphology and good stability for interfacial materials.

The light-harvesting layer, where donor and acceptor materials are blended together to form a so-called bulk heterojunction, acts as the active layer that absorbs solar photons and converts them to free charge carriers in the OSCs. There is no doubt that the development of organic light-harvesting layer materials has played a vital role in improving the PCEs for OSCs from the initial value of 1% to the present level of 18% [Fig. 2(b)].^4,21–47 Fullerene derivatives such as [6,6]-phenyl-C61-butyric acid methyl ester (PCBM) have been widely used as electron acceptors due to their high electron affinity and excellent electron-transport capability.^48,49 A great amount of polymeric or small molecular donor materials have been designed and synthesized to ensure a good match with PCBM in terms of the absorption spectra and molecular energy levels. By using low bandgap polymer donors such as PffBT4T-C₉C₁₃, fullerene-based OSCs have been built that exhibit high PCEs of ∼12%.⁵⁰ However, because of the considerable limitations of fullerene acceptors such as restricted tunability of the electronic/optical properties and large device energy losses, researchers are committed to the gradual development of non-fullerene acceptors. A variety of materials have been synthesized and applied in OSCs, among which acceptor–donor–acceptor (A–D–A)-type NFAs have shown great success. In 2015, a milestone molecule named ITIC was reported, leading to a remarkable improvement in PCE via the use of a PBDB-T polymer as a donor in devices.^40,51 Since then, great efforts have been devoted to tune the optical, electronic, and crystalline properties of NFAs. Currently, Y6 and its derivatives have recorded outstanding PCEs of 15%–18% by matching suitable donor materials in devices due to broader absorption spectra and reduced energy losses.^52–54 

Over the past few years, the PCEs for OSCs have grown very rapidly. As shown in Fig. 2(b), the efficiencies of large-area and eco-compatible processed devices are much lower than that obtained for small-area devices via CB or CF solvents. A lot of work remains to narrow this big gap. Favorable morphology is of critical importance for light-harvesting layers to acquire efficient carrier generation, exciton diffusion, carrier transport, and, thereby, high PCEs.^55–57 On the one hand, processing solvents play a critical role in determining the blend morphology of light-harvesting layers, where bicontinuous interpenetrating networks with favorable nanoscale phase separation are much desired. Halogenated solvents such as CB and CF have been successfully used in most cases mostly because they show very good solubility for materials with highly conjugated structures. However, halogenated solvents face great limitations in large-area preparation as they pose serious environmental problems. Therefore, eco-compatible processability should be taken into account when designing new materials. The median lethal dose (LD₅₀) values (in mg/kg for rats by oral ingestion, available in the Chemical Toxicity Database) for commonly used solvents are shown in Fig. 3.^47,58–69 Note that smaller LD₅₀ values mean higher toxicity. From the LD₅₀ values for solvents, we can find that the commonly used processing solvents for the light-harvesting layers for OSCs such as CB, CF, and 1,2-dichlorobenzene (o-DCB) are highly toxic when compared with nonhalogenated and/or non-aromatic solvents. Aromatic hydrocarbon solvents show an obviously lower toxicity than halogenated organic solvents. The LD₅₀ value for toluene is comparable to that for o-DCB, indicating its high toxicity. Nevertheless, low LD₅₀ values of solvents do not always promise low toxicity. For instance, the LD₅₀ value of methanol is relatively high (5628 mg/kg), and it is believed to be highly poisonous since ingesting a little of methanol may cause permanent blindness or even death. In the mass production of OSCs, the use of low toxicity nonhalogenated and non-aromatic solvents is much desired. To realize the green solvent transition of organic semiconducting materials, Hansen solubility parameters have been applied in fullerene-based OSCs as a practical semi-empirical method for solubility prediction.^70,71 However, the application of Hansen solubility parameters in NFA-based OSCs is rare, and more research is needed to guide solvent processing. On the other hand, different film-forming technologies have varied requirements for materials. When extending the device area by replacing the most widely used spin-coating technique with blade-coating or other printing technologies, it is highly desirable to modify the molecular structures to balance the solubility and crystalline property and, thus, obtain optimal morphology. From the above-mentioned analysis, we can conclude that designing new materials is one of the most feasible strategies for obtaining highly efficient large-area OSC devices via eco-compatible processing solvents.

Highly efficient light-harvesting layer materials should have the following characteristics: (a) broad and strong absorption coverage to match the solar spectrum, (b) suitable molecular energy level alignments between donors and acceptors, (c) a bicontinuous interpenetrating network with appropriate phase separation, and (d) high and balanced carrier mobility. In addition, to fabricate efficient OSCs via eco-compatible solvents, the light-harvesting layer materials should have good film-forming properties and solubility in these solvents.¹⁵ In most cases, to optimize the charge transport properties, highly efficient materials usually have highly rigid structures, resulting in poor solubility in eco-compatible solvents. Therefore, the molecular structures of materials should be finely tuned to realize this goal. Increasing the number of carbon atoms in the flexible side chains and enlarging the dihedral angles between conjugated units may be two useful molecular design strategies. Since the PCEs for NFA-based OSCs are much higher than that for fullerene-based devices, here, in this Perspective we mostly discuss the recent advances for NFAs.

Flexible side chains play a vital role in determining the processability and intermolecular packing properties of the resulting materials.^55,72 Therefore, alkyl side chain engineering has been proven to be one of the most effective strategies for designing light-harvesting layer materials. It is easy to understand that long and branched side chains can improve the solubility of materials in common organic solvents. For instance, NFA Y6 has been used to demonstrate excellent photovoltaic performance. However, it only has good solubility in CB or CF. To improve its solubility in nonhalogenated and/or non-aromatic solvents, we replaced the 2-ethylhexyl (8 carbon atoms) chains in Y6 with longer side chains of 2-butyloctyl (12 carbon atoms) and synthesized BTP-4F-12. The new material shows good solubility in some green solvents, such as o-xylene (o-XY), 1,2,4-trimethylbenzene, and tetrahydrofuran (THF).³⁶ In this case, interestingly, BTP-4F-12 shows enhanced intermolecular stacking compared to Y6 in the in-plane direction, which improves the charge transport. To achieve the goal of fabricating OSCs using eco-compatible solvents, good solubilities for both the donor and the acceptor are required. For highly efficient Y6-based OSCs, the polymer PBDB-TF (also known as PM6) is considered one of the most popular donors. Note that the PBDB-TF polymer is also usually dissolved in o-DCB, CB, or CF and has limited solubility in o-XY, THF, etc. To solve this problem, the chemical structure of PBDB-TF must be modified. In our work, we introduced ester-substituted thiophene units as a third compound in the donor–acceptor alternating conjugated backbone, which can effectively enlarge the molecular twist and, thus, improve the solubility. As a result, we synthesized a ternary polymer named T1 and blended it with BTP-4F-12 for use as the light harvesting layer in the fabrication of OSCs.^36,73 By using THF as the processing solvent, a maximum PCE of 16.1% was recorded for the OSCs, which is comparable to the value obtained for a CF-processed device.

Since the extension of flexible side chains usually results in less ordered intermolecular packing and, thus, decreased charge transport properties, maintaining a suitable morphology is the key to obtaining high efficiencies when conducting molecular modification. For instance, in order to increase the solubility of PBDB-T, Qin et al.⁷⁴ replaced the 2-ethylhexyl group with 2-butyloctyl side chains to design the PBDB-T-BO polymer. This polymer with longer side chains demonstrated improved solubility in THF. However, at the same time, the preaggregation feature disappeared, which is regarded as one of the most likely reasons that this kind of polymer can match well with NFAs to achieve high PCEs. Then, the same group conducted further chemical modification for PBDB-T-BO by using benzothiophene instead of thiophene to extend the conjugated area. After this two-step modification, the PBDB-BzT polymer not only shows good solubility in THF but also possesses a strong solution preaggregation effect. By blending with an NFA named IT-M, the THF processed devices were found to deliver a maximum PCE of 12.1%, which is even higher that the values obtained for CB-based devices. Incorporating siloxane-functionalized side-chains has been proven to be an alternative way to enhance the solubility of the resulting polymer donors. Fan et al. designed and synthesized a wide-bandgap polymer donor PTzBI-Si, which was modified with a siloxane-terminated side-chain.⁷⁵ The absorption spectrum for the PTzBI-Si film shows a complementary absorption and matched energy level alignments with the polymer acceptor N2200. A blend of the two polymers exhibits a high limit of solubility of ∼50 mg/ml in 2-methyl-tetrafuran (MeTHF). In addition, the introduction of siloxane substitutions facilitates the formation of an optimal morphology and a preferential face-on orientation of the polymer chains, leading to well-balanced carrier mobility in devices. As a result, the all-polymer OSC achieved a maximum PCE of 10.1% with a high FF of 0.734, which is a very good result for these kinds of devices. Afterward, the researchers further investigated the application of PTzBI-Si and N2200 in other eco-compatible solvents. By using cyclopentyl methyl ether (CPME) as a processing solvent, the PTzBI-Si:N2200-based OSCs were found to deliver a slightly higher PCE of 11.0% due to increased absorption properties.⁷⁶ In addition, as shown in Fig. 4, all the polymers, FTAZ, PBTA-TF, and PBDTS-TDZ, show good solubility in eco-compatible solvents and achieve good photovoltaic performance in OSCs (summarized in Table I). Furthermore, several studies have been conducted to develop water/ethanol-soluble solar cells via employing hydrophilic oligoethylene glycol side chains modification. Lee et al. developed naphthalene diimide (NDI)-based polymer acceptors with modified oligoethylene glycol side chains, named P(NDIDEG-T).⁷⁷ After blending with the polymer donor PPDT2FBT-A, a PCE of 2.15% was obtained with water/ethanol mixtures as the processing solvent. Additionally, the unencapsulated OSC retained more than 90% of the initial PCE value after 4 days in the air, indicating a remarkable long-term air-stability. However, the PCE of water/alcohol-soluble PSCs lag behind, and the research field is in the early stages. The effective strategies in the material design and device optimization should be developed for the solvent properties of water, which differ from that of typical hydrocarbon solvents.

Designing novel light-harvesting layer materials by increasing the dihedral angles between conjugated units is also an effective method to improve solution processability.^16,50 For instance, perylene diimides (PDIs) and NDIs are classical building blocks for n-type organic semiconductors. These materials have very rigid conjugated backbones, usually resulting in large-sized crystalline domains in the solid state.^78–80 To weaken aggregation effects, linking two or more PDI units to have large dihedral angels has been demonstrated as an effective strategy. For example, N2200 is a very typical polymer acceptor material that is used in field-effect transistors and OSCs, which possesses a thiophene bridge between two NDI units. N2200 can be dissolved well in eco-compatible solvents such as THF and MeTHF, and high PCEs for all-polymer OSCs have been frequently reported.⁸¹ PDI-containing polymer acceptors such as PPDIODT also show good solubility in many eco-compatible solvents due to its twisted conjugated backbone and long side chains. By using the polymer PBDT-TS1 as a donor, as-cast OSCs processed by anisole have recorded a very good PCE of 6.58%.⁸² Recently, Sun et al. incorporated 1.5 mol. % Ir-complex into the polymer donor PBDB-TF. The resulting polymer PM6-Ir1.5 presents enhanced steric hindrance for intermolecular packing and, thus, weakens the tendency for aggregation, as verified by a slightly reduced lamellar diffraction and π-π stacking observed from grazing incidence wide-angle x-ray scattering measurements.³⁷ By blending a Y6-C2 derivative as an acceptor and using toluene as the processing solvent, OSCs have achieved a good PCE of 16.52%.

Spin-coating is a widely used film-forming method in the laboratory owing to its advantages of convenience, low price, and easy-availability. Nevertheless, it is not compatible with mass production as the utilization efficiency of the material is too low, and it is hard to obtain large-area films with a uniform thickness.⁸³ Considerable efforts have been devoted to the transference of the high efficiencies found for spin-coated devices to devices prepared using scalable printing methods, such as blade coating (solution shearing), slot-die coating, and roll-to-roll coating, which are not that straightforward due to the different processing temperatures and film formation mechanisms involved.^84,85 Considering the varied features of such film-forming technologies, modification of the crystalline property and solubility of light-harvesting layer materials is required.

Blade-coating (also called knife-coating) has the advantages of large-area uniformity, a small quantity of material losses, and compatibility with roll-to-roll manufacturing.^14,83 The film thickness can be adjusted by changing the fabrication parameters such as solution concentration, the gap between the blade and the substrate surface, and coating speed. In 2017, the potential for preparing high efficiency OSCs via blade coating with eco-compatible solvents was demonstrated.³⁵ As is known, it is crucial to acquire a suitable morphology for high efficiency OSCs. Solvents with different volatilities have a significant influence on the morphology of OSCs. Zhao et al. investigated the application of solvents with different boiling points for spin-coating and blade-coating.³⁵ A polymer donor PBTA-TF and NFA IT-M were selected as active layer materials, and two eco-compatible solvent-based mixtures, o-xylene/1-phenylnaphthalene (PN) and THF/isopropyl alcohol (IPA), were used as processing solvents. Research shows that the THF/IPA-processed devices prepared by the two above-mentioned coating methods display comparable performance, while the o-xylene/PN-processed devices demonstrate dramatically different PCE values for spin-coating (13.1%) and blade-coating (8.2%). The significant differences in device performance are related to the morphologies of the active layer [Fig. 5(a)], which is influenced by the coating methods and the volatilities of the solvents. When adopting the spin-coating method, a wet film is dried very quickly with the help of rapid airflow and volatility of the solvent. However, the drying process during blade-coating demonstrates an obvious difference between low and high boiling point solvent mixtures. A THF/IPA-processed wet film will be dried solely due to a high saturated vapor pressure, while the drying process for an o-xylene/PN-processed film will be greatly prolonged, leading to over aggregation in the active layer. As a result, the blend films prepared via a spin-coating method show smooth surface topographies and nanoscale phase separation, when using either o-xylene/PN or THF/IPA as the solvent, whereas the o-xylene/PN-processed film prepared by a blade-coating method shows severe phase separation and, thereby, a low PCE value. The low boiling point of THF/IPA allows for a fast drying process and suppressed oversized aggregation, which is beneficial for obtaining a higher PCE. When the device area is scaled up, the morphology control is much more challenging. The bilayer structure of OSCs, where the active layer is composed of coating the solutions of a neat donor and a pure acceptor successively, is favorable for achieving high-performance devices and printing large-area OSCs owing to better charge transport and independent preparation of donor and acceptor materials. Benefitting from the significant spectral overlap between the photoluminescence of the polymer donor and the absorption of NFAs, efficient exciton diffusion in the bilayer heterojunction could be achieved by long-range energy transfer, which is different from fullerene acceptors with less overlap.^86,87 To obtain an optimal morphology, Dong et al. have adopted a sequential bilayer processing method via the use of a mixture of nonhalogenated solvents [Fig. 5(b)]. By using PBDB-TF and IT-4F as the light-harvesting layer material, a PCE of 11.4% was achieved at an area of 1 cm², which is comparable to the PCE value (12.9%) obtained by spin-coating for a small-area device (0.04 cm²).⁸⁸ These results indicate that the bilayer structure is a promising candidate for high performance OSCs distributed with enriched donors on the bottom and acceptors on the top surface, which can also be used for large-area coating/printing technology.

As is well-known, the production of up-scaled roll-to-roll modules is one of the necessary steps from laboratory-scale devices to mass production. Strohm et al. fabricated a semi-transparent module based on P3HT:IDTBR at an effective area of 60 cm² and achieved an efficiency of 4.7% by using the eco-compatible solvent mixture of o-methylanisole.⁸⁹ Zhang et al. developed a polymer donor TPD-3F (Fig. 6). When blending with IT-4F, the efficiencies of devices are insensitive to the processing solvent and technology (spin-coating and blade-coating). A large active-area OSC module (>20 cm²) processed by blade-coating using halogen-free solvents under ambient showed an outstanding PCE of 10.4%.

The inkjet printing process has many advantages such as contactless conformal deposition, compatibility with various substrates, low material usage, and digital control 2D patterning with high resolution.^14,90 Corzo et al. demonstrated the potential transition from the laboratory to factory via the inkjet printing process.⁹¹ The commercially available materials, P3HT and O-IDTBR, were selected as the light-harvesting layer materials. To acquire the inks with optimized process parameters, the viscosity and surface tension of the inks were tuned by adjusting the solvent composition and concentration. With the synergy of ink formulation and optimized process parameters, a PCE of 4.43% was yielded in o-xylene:tetraline (1:1) mixed solvents. Moreover, a 2.2 cm² marine turtle-shaped solar cell was designed and digitally fabricated, indicating the versatility for free-form preparation.

With significant advances of novel materials, the PCEs of OSCs have increased rapidly. However, compared to devices processed by halogenated solvents, the efficiencies for devices processed via eco-compatible solvents lag behind. To realize future applications, replacing toxic organic solvents by eco-compatible solvents to fabricate devices is highly required. Herein, we review the recent progress for eco-compatible solvent-processable OSCs from the aspects of solvent toxicity, material design, and large-area device preparation. In most cases, high efficiency light harvesting layer materials show limited solubility to ensure the best charge transport properties. To solve this problem, the two main strategies used in material design involve optimization of flexible side chains and enlargement of the dihedral angles formed between conjugated units, which are suitable for both donor and acceptor materials. In addition to the new materials, this method is noteworthy for achieving upscale fabrication with suitable and cost-effective means. More studies should focus on eco-compatible solvent-processed OSCs built using suitable fabrication methodologies, such as blade coating and inkjet printing. In addition, some critical issues that need to be resolved include (a) the development of eco-compatible solvent-processable materials with high thickness tolerability (in general, for large-area printing, high active layer thickness tolerability is necessary), (b) the development of film-forming methods that are suitable for large area preparation and reduced PCE loss, and (c) more attention needs to be given to flexible substrates, scalable interlayers, and electrodes that are suitable for large-area production technologies.

This work was supported by the National Key Research and Development Program of China (Grant No. 2019YFA0705900) funded by MOST and the Basic and Applied Basic Research Major Program of Guangdong Province (Grant No. 2019B030302007). The authors acknowledge the financial support from the National Natural Science Foundation of China (Grant Nos. 21835006, 21805287, 22075301, 51961135103, 51961165102, 52003275, and 51673201). H.Y. is a member of the Youth Innovation Promotion Association CAS (Grant No. 2018043). This work was also financed by the Beijing National Laboratory for Molecular Sciences (Grant No. BNLMS-CXXM-201903). Z.G. thanks the financial support from the National Key R&D Program of China (Grant No. 2017YFE0106000) and the National Science Fund for Distinguished Young Scholars (Grant No. 21925506).