Solvent Additives: Key Morphology-Directing Agents for ...download.xuebalib.com/4rm4AGSsHkOv.pdf(KAUST), Saudi Arabia, in 2016. He currently works as a postdoctoral scholar in the

REVIEW

1707114 (1 of 30) © 2018 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

www.advmat.de

Solvent Additives: Key Morphology-Directing Agents for Solution-Processed Organic Solar Cells

Caitlin McDowell, Maged Abdelsamie, Michael F. Toney,* and Guillermo C. Bazan*

DOI: 10.1002/adma.201707114

1. Introduction

Environmental concerns over of the use of fossil fuels, such as air pollution and climate change, drive research and develop-ment in renewable energy sources. In particular, solar energy, in the form of readily available radiant light, has the potential to provide greater power output than the projected global energy demand, even by conservative estimates.[1] This review focuses on photovoltaic technologies based on organic semiconduc-tors that have the advantage of chemical tailoring to optimize

Organic photovoltaics (OPV) have the advantage of possible fabrication by energy-efficient and cost-effective deposition methods, such as solution processing. Solvent additives can provide fine control of the active layer mor-phology of OPVs by influencing film formation during solution processing. As such, solvent additives form a versatile method of experimental control for improving organic solar cell device performance. This review provides a brief history of solution-processed bulk heterojunction OPVs and the advent of solvent additives, putting them into context with other methods available for morphology control. It presents the current understanding of how solvent additives impact various mechanisms of phase separation, enabled by recent advances in in situ morphology characterization. Indeed, understanding solvent additives’ effects on film formation has allowed them to be applied and combined effectively and synergistically to boost OPV performance. Their success as a morphology control strategy has also prompted the use of sol-vent additives in related organic semiconductor technologies. Finally, the role of solvent additives in the development of next-generation OPV active layers is discussed. Despite concerns over their environmental toxicity and role in device instability, solvent additives remain relevant morphological directing agents as research interests evolve toward nonfullerene acceptors, ternary blends, and environmentally sustainable solvents.

Hall of Fame Article

Dr. C. McDowell, Prof. G. C. BazanCenter for Polymers and Organic SolidsDepartments of Chemistry and Biochemistry and MaterialsUniversity of California, Santa BarbaraSanta Barbara, CA 93106, USAE-mail: [email protected]. M. Abdelsamie, Prof. M. F. ToneySLAC National Accelerator Laboratory2575 Sand Hill Road, Building 137, Menlo Park, CA 94025, USAE-mail: [email protected]

The ORCID identification number(s) for the author(s) of this article can be found under https://doi.org/10.1002/adma.201707114.

absorption of the solar spectrum such that thin flexible and lightweight devices can efficiently harvest sunlight. Consequently, they have potential to be integrated directly into the windows of buildings and other architectural structures.[2–4]

Organic photovoltaic (OPV) active layers offer the potential of energy-efficient and cost-effective deposition methods, such as solution processing.[5] OPVs con-sist of blends of electron-donating and electron-accepting polymers or organic small molecules that are cast into thin films from organic solvents and sponta-neously phase separate into small nearly pure domains. However, as described in more detail below, processing from solu-tion opens challenges in that the drying of the solvent greatly impacts the final organization of the blend components in the final photoactive layer. This complex bulk heterojunction (BHJ) morphology of the active layer has a profound impact on the performance of OPVs. Casting from a single solvent usually results in a non-ideal morphology with either insufficient or excessive phase separation between the

BHJ blend components and/or lack of order within domains, thereby reducing the device efficiency. The selection of deposi-tion solvent is constrained by the requirement of solvating the organic semiconductors that comprise the active layer. Thus, additional methods to manipulate the morphology after deposi-tion have been developed, such as thermal annealing and sol-vent vapor annealing. However, including uncommon solvents as additives can affect film formation during deposition, which has distinct advantages compared to postprocessing methods. Fine control over morphological features enabled by such sol-vent additives form a key and versatile experimental handle for optimizing device performance. As the variety of successful solvent additives grows, our understanding of the complex mechanisms underlying the development of the BHJ improves, but we still require deeper insight into these mechanisms to develop predictive design rules.

In this review, we provide a brief introduction to BHJ organic photovoltaics and the advent of solvent additive processing, put-ting it into context with other morphology control strategies. We present the current understanding of how solvent additives impact various mechanisms of phase separation, enabled by recent advances in in situ morphology characterization. Under-standing how such solvent additives affect film formation has

Adv. Mater. 2018, 30, 1707114

http://crossmark.crossref.org/dialog/?doi=10.1002%2Fadma.201707114&domain=pdf&date_stamp=2018-06-13

© 2018 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim1707114 (2 of 30)

www.advmat.dewww.advancedsciencenews.com

prompted their use in other solution-processed organic and hybrid organic–inorganic semiconductor technologies. Finally, we discuss the role of solvent additives in the development of next-generation OPV active layers. Despite concerns over tox-icity and possible role in long term stability, solvent additives will likely remain a relevant morphological control strategy for a variety of optoelectronic technologies that build on the pro-cessability of organic semiconductors, including the develop-ment of less toxic and more benign alternatives.

2. A Brief History of Solvent Additives in Organic Photovoltaics

2.1. Development of the Bulk Heterojunction

Five steps are essential for current production directly from sunlight in any organic photovoltaic device: 1) the active layer absorbs photons, forming an excited electron–hole pair; 2) this “exciton” diffuses to a lower energy state, typically at an inter-face or an impurity; 3) the excitons dissociate into free charges; 4) free electrons and holes diffuse through the active layer; and finally 5), the free charges are collected at the electrodes as cur-rent. OPV active layers typically accomplish this by blending electron-donating (donor, p-type) and electron-accepting (acceptor, n-type) semiconductors. Donors and acceptors have been typically paired based on their energy level offsets (to drive charge separation), absorption of the solar spectrum to maxi-mize photocurrent, and charge carrier mobility to ensure effi-cient transport of free carriers to the electrodes. These organic semiconductors are typically designed to contain π-conjugated aromatic subunits to promote electron delocalization with a minimal fraction of electronically inert side groups needed to ensure solubility. Effective charge generation and transport requires the donor and acceptor components of the film to phase separate into internal domains with high interfacial area and bicontinuous pathways for charge extraction. This delicate organization is known as a BHJ system.

In solution-processed OPVs, the BHJ organization develops primarily as the solvent dries, but the final morphology can be manipulated by the use of solvent additives, postdeposition treatments or both. A possible route to BHJ formation is shown in Figure 1 where donor crystallization is initiated by using a slow-drying poor solvent as an additive. Further phase separa-tion can be promoted by subsequent thermal annealing.

The geometric organization of the BHJ morphology must be tuned in order to take into consideration three properties of organic semiconductors. (1)) Excitons in organic materials are strongly bound. (2)) Bound excitons break up via ultrafast charge transfer to an adjacent molecule whose energy levels are offset relative to the absorbing species, thereby forming free carriers. 3) Excitation lifetimes in organic materials are relatively short such that their diffusion within a domain is lim-ited before recombination. These properties are revealed by key developments in the history of OPV research.

Given their strong absorbance and chemical tunability, organic semiconductors were considered attractive materials for solar cell applications upon their discovery in the late 1970s;[6,7] however, the first observation of significant photovoltaic activity

Caitlin McDowell obtained her B.Sc. in physics and chem-istry from Tufts University in 2008. Working as a laboratory technician with Prof. Timothy Swager at the Massachusetts Institute of Technology sparked her interest in organic electronics. In 2017, she obtained a Ph.D. in materials chemistry under the guidance of Prof. Guillermo C. Bazan,

University of California, Santa Barbara. At the Center for Polymers and Organic Solids, she tracked how molecular structure, solvent choice, and processing additives influ-ence the self-assembly of organic small molecules used in organic photovoltaics. She focused on exploring the novel routes these effects suggest for enhancing performance.

Maged Abdelsamie obtained his B.Sc. in metallurgical and materials engineering from Suez University, Egypt, in 2010. He received M.Sc. and Ph.D. in materials science and engineering under the supervi-sion of Prof. Aram Amassian from King Abdullah University of Science and Technology (KAUST), Saudi Arabia, in 2016. He currently works as a

postdoctoral scholar in the Toney Group at the Stanford Linear Accelerator Laboratory’s Stanford Synchrotron Radiation Lightsource. His research interests include in situ characteri-zation of thin film formation and microstructural analysis of thin films for photovoltaics and optoelectronic applications.

Michael F. Toney received his B.S. from Caltech in 1979 and Ph.D. in surface physics from the University of Washington in 1983. Then, as a NATO Postdoctoral Fellow at Denmark’s Risoe National Laboratory, he used X-ray dif-fraction to study semiconductor surface structures. In 1984, he joined the IBM Research Division in San Jose to focus on

the use of X-ray scattering methods for structure determination for polymer thin films, surfaces, and interfaces. In 2003, he joined the Stanford Synchrotron Radiation Lightsource (SSRL), where he is presently a staff scientist overseeing the SSRL’s X-ray scattering program. He is a pioneer in the use of surface X-ray diffraction for in situ investigations of atomic structure at electrode–electrolyte interfaces in energy storage and of the molecular structure of organic and magnetic thin films.

Adv. Mater. 2018, 30, 1707114



would come later, in 1995.[8] In an organic semiconductor, absorption of light creates an electron–hole pair (i.e., exciton) that is tightly bound (with a binding energy of 0.1–0.5 eV).[9] This contrasts with inorganic semiconductors, whose excitons are loosely bound at room temperature and can thus more readily dissociate into free charges. Few free carriers can be collected at the electrodes for single-component OPV films.[10] For example, PPV and its derivatives have solar cell efficien-cies of 0.1% or lower.[9] Note that the chemical structures of donors and acceptors are provided in Figure 2, while those of solvents, solvent additives, and solid additives are provided in Figure 3.

The strong binding energy of the exciton can be overcome by mixing two phase-separating organic semiconductors with offset energy level alignments, creating a driving force for free carrier generation. In 1992, photoinduced electron transfer between conjugated polymers and fullerenes was discovered.[11] Fullerenes were observed to quench the pho-toluminescence of the polymer in blended films[12] and, later, ultrafast photoinduced electron transfer spectroscopy con-firmed a very fast charge transfer time constant of ≈45 fs.[13] This discovery implied that the efficiency of charge separation could readily approach 100%, compensating for the inherently short lifetime of the strongly bound excitons in π-conjugated materials.

Initial OPV cells were planar heterojunctions, as shown in Figure 4, where the donor and acceptor materials are depos-ited separately to create a well-defined interface for charge transfer.[14,15] This architecture was initially born of conveni-ence: π-conjugated donor materials performed best with a minimum of solubilizing groups and were often polymerized directly onto the substrate.[16] Similarly, unmodified fullerenes also have poor solubility and are better suited for thermal deposition methods. However convenient, the performance of planar architectures is inherently limited by their low interfa-cial area for exciton dissociation.[17] Further limitations arise

from the short excitation lifetimes of organic semiconduc-tors such that excitons typically diffuse up to approximately 10 nm. Thus, ideal planar heterojunction layers should have thicknesses of this order, which limits the absorbance of the active layer.

These properties prompted investigations into 3D BHJ struc-tures, also pictured in Figure 4. In an ideal BHJ, donor and acceptor components form domains on the order of 10–20 nm that are sufficiently interconnected to create charge transport pathways to the electrodes. Achieving this bicontinuous net-work is difficult to template mechanically (unlike the planar junction) so one must rely on self-assembly during solution casting. In practice, BHJ blends are cast from a common sol-vent, relying on the different interaction energies of the com-ponents to promote phase separation. This is particularly true for high molecular weight polymers and the soluble fullerene acceptors that have a small free energy of mixing.[18] Addition-ally, crystallization into pure ordered domains provides addi-tional driving force for phase separation, particularly for more rigid small molecule donors.

As mentioned previously, the morphology of the BHJ active layer is a strong determinant of device performance.[19] As an example, Figure 5a presents a simplified energy level diagram for a conventional p-DTS(FBTTh2)2:PC71BM device, empha-sizing where photovoltaic processes occur within the BHJ morphology in Figure 5b. While excitons can form and dif-fuse within the film, they have little driving force to separate into free charges except near the interface between donor and acceptor domains—energetically, this interfacial position leads to a charge-transfer (CT) state. The energy level of the CT state is determined by its local environment such that states in inti-mately mixed phases are lower in energy.[20] This implies that the interface between donor and acceptor domains need not be sharp—in fact, a concentration gradient between the two components at an interface has been suggested to create an energetic gradient that promotes efficient charge separation.[21]

Adv. Mater. 2018, 30, 1707114

Figure 1. A schematic showing a possible route to formation of the BHJ morphology during deposition, using a solvent additive to extend the drying time and postdeposition thermal annealing to manipulate the final morphology. The diagram is based on in situ studies of p-DTS(FBTTh2)2:PC71BM cast from CB with DIO as a solvent additive.[92] Adapted with permission.[191] Copyright 2017, Elsevier.



Optimal solar cell performance can be achieved by balancing ordered and pure domains of each component with an inter-facial mixed phase to promote exciton dissociation into free charges.[22,23] This three-phase morphology forms the pre-sent basis of our understanding of the most desirable BHJ

structure—however, the optimal ratio of these three phases often depends on the specifics of the blend.

Six key components determine the BHJ morphology and thus OPV device performance: molecular structure, the primary casting solvent (determined by the solubility of the components),

Adv. Mater. 2018, 30, 1707114

Figure 2. Chemical structures of the polymer donors (blue), small molecule donors (red), and acceptors (black) discussed in this review.



the ratio of the components, overall concentration of compo-nents, solvent (or other) additives, and processing conditions (both during and after deposition). Solubility of each compo-nent must be sufficiently high to create a blended film of thick-ness that effectively harvests light. For optimum efficiency, a deposition solvent must simultaneously dissolve the donor and acceptor components but also promote their phase separation into bicontinuous networks. Finding a solvent that effectively dissolves both components has historically limited selection

largely to halogenated or aromatic solvents like chloroform (CF), chlorobenzene (CB), and ortho-dichlorobenzene (oDCB). The choice of deposition solvent also determines the drying time and kinetics of this process, and the wetting behavior of the solvent and BHJ components on the desired substrate. In practice, these requirements are difficult for a single solvent to meet simultaneously, which is why many blends are kineti-cally trapped in intimately mixed phases as-cast. Additional pro-cessing, during or after deposition, can allow the morphology to approach a more efficient phase-separated BHJ and thus improve performance.

The figures of merit for an OPV device are discussed below, highlighting how the BHJ morphology contributes to each metric.[24,25] They are also illustrated in Figure 6 for a blend of a small molecule donor p-DTS(FBTTh2)2 with PC71BM cast from CB. The morphology of this donor:acceptor pair is strongly impacted by the presence of the 1,8-diiodooctane (DIO) solvent additive, boosting OPV efficiency from 1.8% (blue traces) to 7.0% (green traces).

Short-circuit current, JSC: The current flow when the device is shorted and no bias is applied.

Free charges move due to the internal field established by different contact work functions. The JSC depends on the number of photons absorbed (due to the materials’ absorp-tion profile and extinction coefficient) and the efficiencies of the exciton separation and charge transport processes. BHJ

Adv. Mater. 2018, 30, 1707114

Figure 3. Chemical structures of the deposition solvents (red), aromatic solvent additives (blue), nonaromatic solvent additives (green), and solid additives (black) discussed in this review.

Figure 4. Schematic of common OPV device architectures.



morphologies with ordered domains connected by percolat-ing pathways greatly increase these efficiencies and thus JSC.

Open-circuit voltage, VOC: The bias required to reduce the cur-rent flow to zero, effectively countering the built-in field.

This voltage reflects the difference in the quasi-Fermi levels of the blend components, estimated by the offset between the highest occupied molecular orbital of the donor and lowest unoccupied molecular orbital of the acceptor.[26] Morphology optimization can improve VOC by reducing recombination losses but only to the limit pinned by the quasi-Fermi levels.[27]

Fill factor, FF: A measure of the cell’s deviation from ide-ality due to competition between charge collection and recombination.[28,29]

This is determined by the ratio of the maximum measured power (light gray) versus the power produced by an ideal diode (dark gray). Like JSC, this term depends on efficiency of

charge generation and transport processes.

Power conversion efficiency, PCE: The maximum power output.

A measure of how much electrical power is generated relative to the incident solar spectrum (AM1.5), PCE is related to the three factors discussed above via the following equation

PCE % FF /SC OC inputJ V P( )( ) = (1)

External quantum efficiency, EQE: The wavelength depend-ence of the JSC.

EQE measures the charges collected relative to the number of incident photons, as a function of wavelength and with-out bias applied. Thus, the integrated area under the curve should equal the JSC. Comparing the EQE spectra to the ab-sorbance profiles of the donor and acceptor can deconvolute their relative contributions to the photocurrent.

Adv. Mater. 2018, 30, 1707114

Figure 5. a) Energy level diagram for an OPV device with a p-DTS(FBTTh2)2:PC71BM[58] active layer in conventional architecture. b) Mechanism of energy production in BHJ OPVs: 1) photoabsorption to create an exciton, 2) exciton diffusion to an interfacial CT state, 3) exciton dissociation into free charge carriers, 4) charge transport of electron and hole through bicontinuous network, and 5) charge collection at the electrodes.

Figure 6. Relevant metrics for assessing OPV performance, including a) power curve, b) current–voltage plot, and c) external quantum efficiency spectra. The data pictured are for a small molecule donor p-DTS(FBTTh2)2 blended with PC71BM cast from CB, with and without DIO as a solvent additive to boost performance. Adapted with permission.[58] Copyright 2012, Wiley-VCH Verlag GmbH & Co. KGaA.



Internal quantum efficiency, IQE: The EQE normalized by the light absorption efficiency.

IQE (not pictured) measures the charges collected relative to the photons absorbed by the film and is thus by definition higher than EQE.[22,30] It can approach 90% for optimized blend morphologies, indicating low rates of charge recombi-nation.[20]

An important innovation in processing was thermal annealing, a postprocessing technique found to be particu-larly effective for P3HT:PC61BM. Regioregular P3HT is capable of forming highly crystalline domains but the pres-ence of PC61BM kinetically hinders the donor from ordering. In 2003, Padinger et al. studied how heating the film after casting allowed the BHJ to evolve toward a morphology with improved device performance in significant detail.[31] Thermal annealing of the film softens the P3HT polymer matrix, which allows PC61BM to diffuse out of disordered/amorphous P3HT domains to form PC61BM-rich aggregates. Amorphous P3HT domains, now fullerene free, can readily align into crystalline domains. The result is a three-phase morphology composed of ordered donor and acceptor regions that are surrounded by amorphous mixed phase, as previous described. The evolution stages of the BHJ morphology during thermal annealing[32] and solvent annealing[33] were later confirmed and character-ized in greater detail using grazing-incidence X-ray scattering techniques.[34]

Swelling the film with solvent vapor can also be used to reverse vitrification of the blend, allowing the film to approach a more thermodynamically favored morphology. In 2005, Li et al. improved the PCE of P3HT:PC61BM from 3.3% to 4.4% by placing the as-cast film in a closed container directly after depo-sition, slowing the rate of oDCB evaporation.[35,36] Solvent vapor annealing (SVA) is an extension of this technique by intro-ducing a media reservoir inside a closed container—allowing for swelling with alternate media. SVA during casting is also possible.[37] SVA differs from the “slow growth” described by Li et al. in that it provides enough molecular mobility for local rearrangements without fully dissolving the components.[38] Similar to thermal annealing, SVA softens the polymer matrix and allows the BHJ components to phase separate. However, unlike thermal annealing, the choice of solvent vapor can guide the phase evolution toward larger or finer features, depending on the solvents affinity for the BHJ components and its vapor pressure.[39]

Both thermal annealing and solvent vapor annealing drive the system toward the most thermodynamically stable state, which may not be the optimal arrangement for photovoltaic perfor-mance. The morphologies accessed by postprocessing may also be limited by the nanostructure of the initial film formed during solution casting. Thus, extreme morphologies such as nearly homogenous blends or overly separated blends have difficulty achieving the ideal degree of phase separation via postprocessing despite careful optimization efforts. Strategies were therefore sought that could template a structural arrangement closer to the desired morphology during the timescale of film formation. The deposition solvent was noted to have a strong influence morphology but choice of deposition solvent is strongly limited by the solubility of both donor and acceptor components.

These observations led to the idea of a solvent additive, a small volume of additional solvent or antisolvent (less than 10%) that would have a disproportionately large impact on the drying and film formation process. This approach is distinct from a cosol-vent (>10%) as cosolvents are limited by the same parameters as the primary solvent choice, although both can be used stra-tegically to similar ends. One significant advantage of a solvent additive is the relative freedom in its selection; at a minimum, it must be partly miscible in the deposition solvent but is free to solvate the donor, acceptor, both or neither. Solvent addi-tives affect the BHJ microstructure by manipulating the molec-ular order and orientation of pure donor/acceptor domains and their degree of phase separation. Often, solvent additives improve the overall crystalline order, but the effect of any given additive varies from one donor:acceptor pair to another. In the next section, we discuss the history of solvent additive usage in OPVs and criteria for their selection.

2.2. Advent of Solvent Additive Processing

The first successful solvent additive was reported in 2006. Peet et al. initially observed higher photocurrents from P3HT:PC61BM blend films when excess n-octylthiol (1-octa-nethiol) was used to incorporate gold nanoparticles for plas-monic resonance enhancement. The n-octylthiol itself was serendipitously found to induce structural order in P3HT, increasing mobility in BHJ thin film transistors similar to thermal annealing.[40] In a follow-up paper, Peet et al. found that alkyl dithiols of varied length (from propyl to octyl) also boosted the efficiency of PCPDTBT:PC71BM, a blend that did not significantly improve with thermal or solvent vapor annealing.[41] Unlike the case with P3HT:PC61BM, the alkyl dithiol additives primarily promoted phase separation, with minimal impact on overall donor crystallinity. The greatest performance enhancement occurred for octanedithiol (ODT), which boosted PCEs from 2.8% to 5.5%. X-ray dif-fraction studies by Rogers et al. also confirmed that these solvent additives do subtly increase the crystalline correla-tion lengths of the donor’s features, indicating more perfect crystallites.[42]

A pivotal paper by Lee et al. identified potential selection criteria for solvent additives based on their chemical proper-ties, demonstrated with the same system, PCPDTBT:PC71BM. Notably, this blend does not phase separate significantly when processed directly from CB, leading to a PCE of 3.4%. First, Lee et al. demonstrated that ODT does not react or dope the polymer in any way. The morphological changes and improved performance induced by ODT persist after residual solvent is removed from the film by high vacuum—thus, ODT is purely a “processing additive.”[43] Also, ODT selectively dis-solves PC71BM from the active layer when PCPDTBT:PC71BM films are dipped in the additive, leaving the polymer matrix behind as illustrated in Figure 7.[44] When used as a solvent additive, ODT remains in the film after the deposition sol-vent evaporates due to its lower vapor pressure. The slower drying of ODT promotes the formation of three phases: 1) an aggregated polymer phase formed by the rapid evapora-tion of the deposition solvent; 2) a mixed polymer–fullerene

Adv. Mater. 2018, 30, 1707114



phase; and 3) fullerene solvated by ODT, which becomes fullerene aggregates as the additive evaporates. Again, this “three-phase” morphology is believed to drive efficient exciton dissociation.[21,23] Lee et al. proposed that processing additives should be selected based on two primary criteria: 1) selective solubility of the fullerene component; and 2) having a higher boiling point (and thus lower vapor pressure) than the deposi-tion solvent.[43]

Further investigations probed the impact of six 1,8-di(R)octane compounds (where R is thio-, chloro-, bromo-, iodo-, cyano-, and acetate) that fit this criterion on the morphology

of PCPDTBT:PC71BM. All six additives pro-moted greater phase separation of the pol-ymer and fullerene at 2.5% v/v additive con-centration but only three (thio-, bromo- and iodo-) led to performance increases, to a max-imum PCE of 5.1% for 1,8-diiodooctane.[43] At this concentration, the other three addi-tives lead to overcrystallization, and thus poor charge extraction and low JSC. Figure 7 shows the internal polymer networks exposed when ODT is used to wash away PC71BM. Note that larger and more hierarchical structures are formed when processed with ODT (right side). Hierarchical structures are created by the evaporation pathway of residual solvent and are proposed to be beneficial for charge percolation without sacrificing interfacial area.[45,46]

A study by Peet et al. correlated changes in the PCPDTBT:PC71BM morphology to shifts in BHJ film absorbance with redshifts indicating more ordered films.[47] This study confirmed that the evaporation of the solvent additive impacts the evolution of ordered structures, as shown in Figure 8. Little changed in absorbance during the evapora-tion of the deposition solvent trichloroben-zene (TCB) (in black) but the traces redshift over the longer timescale of DIO evaporation. These observations led to the general conclu-sion that the critical morphological reorgani-zation takes place during the timescale of DIO evaporation.

Both ODT and DIO proved to be effective solvent additives for BHJ blends that do not adequately phase separate during deposi-tion, by extended drying times and inducing donor crystallinity as poor solvents. How-ever, as a broader diversity of donor mate-rials became available, the number of BHJ blends that did not respond well to these two solvent additive treatments increased. This was the case for a novel copolymer of dithienosilole and benzoxadiazole, P1, which formed micrometer-sized domains when processed with PC71BM due to its poor solubility in the deposition solvent, CB.[48] Hoven et al. selected 1-chloronaph-thalene (1CN) as a potential solvent addi-

tive because of its high boiling point and its ability to solvate the polymer to a greater extent than the deposition solvent CB. Using 2% v/v 1CN, the domain sizes of the BHJ com-ponents decreased dramatically and PCE values increased from 1.6% to 4.9%.[48] 1CN creates a finer BHJ morphology by promoting a more intimate arrangement of donors and acceptors that are otherwise prone to rapid demixing.[49] Other aromatic solvent additives, like diphenyl ether (DPE), also appear to increase donor and acceptor miscibility.[50,51] In particular, Choi et al. used DPE to optimize the morphology

Adv. Mater. 2018, 30, 1707114

Figure 7. AFM and TEM images of BHJ films cast from PCPCTBT:PC71BM without and with ODT. AFM image of BHJ film a) without and b) with ODT; AFM image of PCPDTBT networks after removal of PC71BM in BHJ film c) without and d) with ODT; and TEM image of exposed polymer networks e) without and f) with ODT. Reproduced with permission.[43] Copyright 2008, American Chemical Society.



of DT-PDPP2T-TT:PC71BM in films with a thickness of up to 300 nm, thereby improving the PCE from 3.2% to a remark-able 9.5%.[52] A recent comprehensive study on the effects of these solvent additives (ODT, DIO, 1CN, and DPE) showed that DPE most readily creates optimized BHJ morpholo-gies for five key polymer:fullerene blends (including P3HT, PCDTBT, and PTB7).[53] The researchers attribute this to DPE’s favorable interactions with the conjugated backbones of the polymers as a “theta” solvent, wherein polymers behave as ideal solvated chains.[53] Indeed, 1CN has also been used as deposition solvent for similar polymers to successfully fab-ricate thin-film transistors with comparable performance to films cast from CB.[54]

Thus far, we have discussed the impact of solvent additives on the morphologies of polymer-based BHJ films but they have similar (if not more pronounced) effects on blends that contain molecular donors. Molecular donors have defined structures without batch-to-batch variability[55,56] and tend to exhibit higher crystallinity than their polymer counterparts. However, their self-assembly can be kinetically hindered by the presence of an acceptor, leading to poor phase separa-tion. This is readily ameliorated by solvent additives: Sun et al. increased the PCE of p-DTS(PTTh2)2:PC71BM from 4.5% to 6.7% with just 0.25% v/v DIO in CB,[57] and van der Poll et al. used 0.4% v/v DIO to achieve 7.0% PCE (compared to 1.8% as-cast) for p-DTS (FBTTh2)2:PC71BM.[58] These optimized PCEs demonstrated that molecular donors could compete with their polymer counterparts, provided that an optimal blend morphology is reached. Blends with small molecules that are prone to overcrystallization also respond to solvent additives that prevent early phase separation (i.e., compatibilizers), like 1CN.[59,60]

Notably, the morphologies of small molecule donor:fullerene blends are more sensitive to additive concentration, leading to very effective changes in performance metrics. Blends with molecular donors also generally optimize at lower additive concentrations, typically less than 1% v/v. Rationales based on more facile molecular diffusion, lower polydispersity, and less viscous deposition solutions leading to lower solvent reten-tion have been suggested. However, this observation may also stem from defining the solvent additive concentration relative

to the deposition solvent rather than the BHJ components themselves; Love et al. advocate reporting solvent additives as a mole ratio relative to the BHJ components rather than as a percentage of the solvent.[61] Low vapor pressure selectively solvating additives arguably behave more like additional sol-utes rather than modifiers of the solution properties. For a small molecule system of p-SIDT(FBTTh2)2:PC71BM in CB, Love et al. found the optimal concentration of DIO scales linearly with the BHJ component concentration over a wide range, from 20 to 80 mg mL−1, but the molar ratio of DIO to BHJ components remains constant at 0.38.[61] The DIO con-tent of related donors p-DTS(PTTh2)2 and p-DTS(FBTTh2)2 (as discussed above) also optimizes at this molar ratio.[57,62] This direct stoichiometric relationship between the number of sol-vent additive molecules and BHJ components makes intuitive sense for molecular blends; for larger polymer chains, higher solvent additive concentrations are required to reach the same degree of molecular interaction per conjugated subunit. Thus, while polymer BHJ blends tend to have lower polymer con-tent, with smaller D:A ratios and lower overall concentrations, they yield thicker films with a greater number of conjugated subunits and thus require more additive to develop their morphology.

A broad search for novel solvent additives began using prominent additives (e.g., ODT, DIO, and 1CN) as guides. For example, varied lengths of alkyl dithiols,[41,63] alkyl dihalides[64] and alkyl diols[65] exhibit similar effects on BHJ morphologies as ODT and DIO, but the PCEs provided by their optimal addi-tive concentrations differ. 2-Bromonaphthalene[66] and 1-meth-ylnaphthalene[67] increase miscibility of the BHJ components in a similar manner to 1CN. A pentafluorobenzene derivative of DPE[68] and perfluorinated DIO[69] show greater affinity for fluorinated polymers, increasing their surfactant characteris-tics and thus morphological impact. Though these examples are far from exhaustive, additives with similar chemical struc-tures can clearly access the same enhancement mechanisms but due to their size or the nature of their electron-rich sub-unit, some solvent additives are better suited for particular BHJ blends.

Effective additives have also been found empirically by spe-cifically looking for exceptions to the low vapor pressure and selective solubility “rules-of-thumb.” For example, Chu et al. used dimethyl formamide (and DMSO) to boost the perfor-mance of PCDTBT:PC71BM from 6.0% to 6.7% (and 7.1%) despite these additives’ comparable vapor pressures to the deposition solvent, oDCB.[70] Similarly, Mahadevapuram et al. found that rapid evaporation of the poor solvent THF nucle-ates crystallization in P3HT:PC61BM that is still swollen with oDCB, templating the hierarchical growth of the bicontinuous BHJ network.[71] THF has also been effective for solvent vapor annealing postdeposition, capable of swelling and softening BHJ films presumably due to specific intermolecular inter-actions;[39] solvent swelling may also be at play when THF is used as a solvent additive.

Other types of additives have been discovered. Graham et al. reported in 2011 that a polymeric oil, PDMS, increased the PCE of a blend of an isoindigo-based small molecule il(TT)2 and PC71BM from 1.3 to 2.2%, even at a low 0.05 mg mL−1 concentration, by nucleating smaller donor domains.[72]

Adv. Mater. 2018, 30, 1707114

Figure 8. Absorption spectra of PCPDTBT:PC71BM in the a) 250–300 nm range and b) 600–900 nm range as a function of time after spin coating for 10 s at 2000 rpm from TCB with 2% DIO in 90 s intervals (black and blue) and then 10 min intervals (red). Reproduced with permission.[47] Copyright 2008, American Chemical Society.



Unlike previous additives, this macromolecule remains in the film indefinitely after casting so its morphological impact cannot be driven by evaporation. Instead, PDMS more likely operates as nucleation agent. As a common lubricant, PDMS contamination may factor into the performance variability of solution-processed BHJ OPVs deposited from plastic syringes.[73,74]

Systematic computational approaches to discovering new additives have been proposed; the morphology of a multicom-ponent organic film like the BHJ is correlated to the degrees of interaction among the various active materials, solvents, addi-tives, and interfaces.[75] Hansen developed a methodology that splits interfacial energies into dispersive, polar, and hydrogen-bonding components. Hansen solubility parameters (HSPs) for solids and solutes can be characterized independently and compared to predict the solubilities, miscibilities, and wetting behavior of potential blend combinations.[76–78] This method is well established in the polymer coating and ink formulation industries.[79] In 2004, Hansen and Smith presented the solu-bility parameters and miscibility of C60 with a broad selection of solvents and common nonconjugated polymers,[80] which helped identify potential solvents and additives for fullerene-based OPVs. Walker et al. adapted their methodology to iden-tify deposition solvents for a small molecule DPP(TBFu)2 and PC71BM in 2011, which form intimately mixed films from CF. HSPs predicted carbon disulfide, thiophene, and trichloroethylene may be better deposition solvents than CF (from a selection of 24), but only carbon disulfide showed the desired phase separation as-cast.[81] All solvents tested showed similar morphology and performance to CF upon annealing, however.

Shortly after, Graham et al. used HSPs to choose seven sol-vent additives with various levels of solubility for il(TT)2 and PC61BM.[82] Poor solvents (e.g., PDMS, HD, DEG-DBE) for both components nucleate more donor crystallites and create smaller domains than films cast from CB, increasing JSC, FF, and hole mobility. Good solvents for the polymer (e.g., NMP, DIO, and 1CN) exhibited lower PCEs and mobilities due to the overgrowth of domains. This is illustrated in transmission electron micros-copy (TEM) images of the blend film morphologies in Figure 9, presented with hole mobilities and JSC values. These “good sol-vents” that yield poor performance are common processing addi-tives that often lead to enhanced performance in other systems, emphasizing that additive processing must be tailored to the indi-vidual material system. Graham et al. also found that the polar additive TEG increased the VOC by modifying the work function of PEDOT:PSS, but did not significantly change the BHJ mor-phology.[82] This study further demonstrated that additives with orthogonal mechanisms may be used simultaneously to syner-gistically enhance performance, such as manipulating the mor-phology and contact resistance with PDMS and TEG, respectively.

The diversity of chemical structures and behaviors within the growing catalog of successful solvent additives makes classification challenging. Solvent additives are often simply classified into nonconjugated or aromatic structures since examples from each category tend to have the same effects on a given system.[83,84] Nonconjugated solvent additives, such as ODT and DIO, tend to increase crystallinity, possibly due to increased affinity for the alkyl side chains; aromatic solvent additives, such as 1CN and DPE, have higher affinity for the conjugated backbones of OSC materials, tending to increase their miscibility with fullerene.[85] However, this simplistic

Adv. Mater. 2018, 30, 1707114

Figure 9. Top down bright field TEM images of il(TT)2:PC61BM cells with 1.0 mg mL−1 of solvent additive after 100 °C thermal annealing, with μh (cm2 V−1 s−1) and JSC values indicated at the bottom of each image. The darker regions are PC61BM rich while the lighter regions are donor rich. Adapted with permission.[82] Copyright 2012, Wiley-VCH Verlag GmbH & Co. KGaA.



division does not provide insight into variations in efficacy for these additives with different BHJ blends and deposition solvents.

Machui et al. expanded the criteria proposed by Lee et al.[43] into a four category classification system based on a matrix of selective solubility of the donor (or acceptor) and the evapo-ration kinetics due to the vapor pressure of the additive.[86] The researchers tested the effect of potential additives from each category (as shown in Figure 10) on the performance of P3HT:PC61BM blend films, processed with up to 30% v/v of additive in CB. Additives with high volatility, e.g., acetone and benzene, have little impact on the final BHJ morphology as long as the components remain well solvated. Volatile addi-tives evaporate almost entirely during deposition such that the slower drying kinetics of the deposition solvent is largely responsible for the microstructure of the active layer. Thus, these volatile additives can be used to optimize properties of the deposition solution in ink formulation, such as wetting, surface tension, and viscosity. Poor solvents with low volatility (e.g., propylene carbonate) lower performance considerably, even at very low concentrations. The lingering poor solvent accelerates aggregation of the BHJ components and serves an insulating contaminant.[86] The majority of successful solvent additives belong to the fourth category, with both low volatility and selec-tive solubility, e.g., bromoanisole. The slow evaporation of these additives leads to an enriched microstructure by solvating the fullerene as the donor self-assembles.[86]

In the classification system outlined by Machui et al., solvent additives can be ranked within each category, creating a 2D matrix. For example, both bromoanisole and ODT selectively solvate PC61BM to a similar degree, but ODT has a lower vapor pressure. ODT thus dries more slowly and can lead to

the overgrowth of domains at high concentrations. Bromoani-sole dries relatively faster, and thus the BHJ morphology is less sensitive to the concentration of this additive.[86] This also explains the difference in the effects of nonconjugated and aromatic additives, such as ODT and 1CN, even though both belong to the “low volatility, selective solubility” category. While aromatic additives still solvate fullerenes more than the donor component, the relative degree of selectivity is less than with nonconjugated additives. These updated criteria and matrix designations account for the solvent additive’s specific interac-tions with the blend components and deposition solvent, which can explain the behavior variation observed for a given addi-tive. Consider how DIO and 1CN were observed to diminish the performance of il(TT)2:PC61BM:[82] as good solvents for the donor il(TT)2, DIO, and 1CN score lower in selective solvation for this blend than for other systems, like P3HT:PC61BM. Such understanding enables adaptive additive selection: e.g., the nonfullerene acceptor PDI aggregates in large domains when PTB7:PDI is processed from slow drying CB, but this behavior can be suppressed with 0.4% v/v 1CN additive. Switching the deposition solvent to volatile CF also reduces PDI domain size but does not provide sufficient time for PTB7 to organize, which can be remedied with 0.4% v/v DIO.[87]

3. Mechanistic Understanding of Solvent Additive Effects on BHJ Morphology

In the next section, we discuss the current understanding of the different enhancement mechanisms afforded by solvent additives. By emphasizing how solvent additives impact the formation of the BHJ during processing, it should be possible

Adv. Mater. 2018, 30, 1707114

Figure 10. Left: Categorization of additives according to solubility and volatility. Right: Device characteristics of different additives: a) open-circuit voltage (VOC), b) short-circuit current density (JSC), c) FF, and d) PCE. Adapted with permission.[86] Copyright 2015, Wiley-VCH Verlag GmbH & Co. KGaA.



to move toward more proscriptive guidelines for choosing and combining them rather than previous descriptive approaches. Mechanistic insight would not be possible without tools for directly characterizing the BHJ morphology as OPV perfor-mance metrics do not provide sufficient information on their own. Optical absorbance, X-ray scattering techniques, atomic-force microscopy (AFM), and TEM tomography have been invaluable in shaping our understanding of structure–mor-phology–performance relationships.[24] Now that these tech-niques have been adapted to provide in situ measurements, we have the ability to gain insight into when and where sol-vent additives influence the film formation process toward more effective BHJ morphologies. As such, we highlight in this review the use of in situ measurement techniques that provide mechanistic insight into the film formation process during deposition.

Before discussing particular BHJ blends, it is worth reiter-ating that the identity of the donor, acceptor, and deposition sol-vent must be taken into account when considering the impact of solvent additives: their self-assembly tendencies determine the baseline BHJ morphology and as-cast performance. Thus, the mechanisms of solvent additives are understood largely through in-depth studies of donor:acceptor pairs whose BHJ morphology is already well characterized. Regioregular P3HT is one of the most widely studied donor polymers, capable of forming highly crystalline domains. Its performance is sensi-tive to the size of these domains and thus to a variety of pro-cessing methods. However, P3HT is a homopolymer, unlike the higher performance donor–acceptor copolymers used today. While much has been learned from studying P3HT, conclusions drawn from this polymer donor may not be gen-erally applicable.[88] Influential work has been done using the donor–acceptor polymer PCPDTBT and its many derivatives,[41] but the current morphological standard is PTB7 and its struc-tural derivatives.[89] PTB7 is both commercially available and sensitive to processing, and thus a good candidate for thor-ough morphological study; in particular, its higher optimized PCE provides a wider window for performance variation (ranging from 5% to 9%) from which insight is drawn. Con-sidering small molecule donors, early performance successes for solvent additives were achieved for p-DTS(PTTh2)2

[57] and p-DTS(FBTTh2)2

[62] and subsequent morphological studies of these molecules were influential.[90–93] However, there is greater structural variety among small molecule architectures and thus less agreement on a representative model material for morpho-logical studies. Finally, in situ studies on the mechanisms of solvent additives to date involve functionalized fullerene deriva-tives such as PC61BM and PC71BM. As such, solvent additive selection criteria were initially suggested based specifically on selective solvation of the fullerene acceptor,[43] although solva-tion of either component may be sufficient.

3.1. Thermodynamic and Kinetic Factors in Film Formation

The formation of the BHJ in a multicomponent solvent system is too complex to easily explain by simple mechanisms. First, we must establish the thermodynamic and kinetic framework that guides the formation of the BHJ when casting from a

single solvent. From there, we can understand how solvent additives can be used to influence the thermodynamics and kinetics of film formation. The main parameters that control BHJ formation during solution processing involve: 1) kinetic parameters, such as the vapor pressure of the solvents, and the deposition conditions that collectively define the drying kinetics of the mixture; 2) thermodynamic parameters, such as the solu-bility of donor and acceptor materials in the solvent(s), their ease of crystallization/aggregation, and the mutual interactions between the solvents and the donor and acceptor solutes.

One method for predicting the solidification pathway from a homogenous solution models the phase separation pro-cess using the homogeneous Flory–Huggins free energies of the multicomponent blend.[94–97] So far, such models have only been rigorously applied to three component systems—solvent:donor:acceptor (S:D:A) mixtures—but this methodology can be extended to include more components.[95,97] Expressions in the Flory–Huggins equation account for free energy contri-butions that arise from interfacial interactions between locally separated phases; e.g., the expression for a three-component system of solvent, acceptor, and donor is given by

/ ln( ) / ln( ) / ln( )S S D D D A A A

SD S D SA S A DA D A

G RT N Nϕ ϕ ϕ ϕ ϕ ϕχ ϕ ϕ χ ϕ ϕ χ ϕ ϕ

( ) ( )= + ++ + +

(2)

where ϕi is the volume fraction of component i, χij is the Flory–Huggins interaction parameter between components i and j, and Ni is related to the degree of polymerization and can be estimated by the number of lattice segments constituting one molecule of component i. Equation (2) can used to calculate a ternary phase diagram of S:D:A mixture that can predict the phase behavior of the ternary blend. An example is provided in Figure 11a and construction of such ternary phase diagrams are explained elsewhere in the literature.[94–97]

In Figure 11b, we summarize the main phase separation processes from an initially homogenous liquid phase. During solvent evaporation, two competing processes are driven by the change in the mixture’s composition: liquid–liquid (L–L) or solid–liquid (S–L) phase separation. L–L demixing begins upon crossing from the one-phase to the two-phase region via spinodal decomposition, while S–L demixing occurs when either the donor or acceptor component reaches its solubility limit in the deposition solvent. Depending on the system, the two demixing processes may occur simultaneously or one may precede the other. If neither process occurs, the mixture will end up with an intimately mixed blend morphology.

A schematic representing the ternary phase diagram of solvent:donor:acceptor mixtures with different donor–acceptor interaction parameters (χDA) are shown in Figure 11a. The composition space in the ternary phase diagram is divided into a single phase region where the blend mixture is stable, a two-phase region of instability surrounded by the spinodal curve (solid), and a metastable region between spinodal (solid) and binodal (dashed) curves. The dashed black line in Figure 11a a represents the changes in composition during solvent evapo-ration. Starting from an initially homogenous single phase, L–L demixing begins upon crossing from the one-phase to the two-phase regions. The onset of L–L demixing strongly corre-lates with the interaction parameter between the donor and the

Adv. Mater. 2018, 30, 1707114



acceptor, χDA.[94,95] A donor and acceptor pair with low compat-ibility (high χDA) exhibit a large two-phase region. As a conse-quence, the composition crosses into the unstable region when the mixture is rich in solvent content, as shown by the blue curves in Figure 11a. Under these circumstances, L–L demixing leads to overcoarsening of donor and acceptor domains, driven by the strong repulsive interactions of the donor and acceptor and facilitated by the high solvent content.

Another factor that affects the extent of L–L demixing is the depth to which the blend enters the two-phase region during the deposition process (known as quench depth); fast drying kinetics with deep quench depth are associated with large scale phase separation, while slow evaporation reduces the extent of L–L demixing.[95,98] On the other hand, a donor and acceptor pair with high compatibility (low χDA) has a large one-phase region; thus, the mixture composition reaches the instability region later during solution casting, when there is low solvent content (see red curves in Figure 11a). In this case, L–L phase separation is limited (if not prevented) by the weak repulsive interactions between the donor and acceptor molecules and their low mobility within the film due to low solvent content. Moreover, the solutes may have already reached supersatu-ration such that S–L phase separation drives film formation before significant spinodal decomposition can occur.

S–L demixing occurs when either donor or acceptor begins aggregating after reaching its solubility limit or supersatura-tion in the deposition solvent. The progress of S–L phase sepa-ration is governed by the thermodynamics and kinetics of the nucleation and growth processes of the aggregated solid phase. Materials with a strong tendency to crystallize (i.e., having a low nucleation barrier) tend to segregate easily in solution, unlike materials that resist crystallization (i.e., having a high activation barrier).[93,99] Competition between the kinetics of nucleation and growth of the aggregated solid phase and the solvent drying

kinetics also affect the progress of S–L demixing; slow evapo-ration favors prolonged aggregation or crystallization and thus increases the degree of S–L demixing. Importantly, S–L phase separation competes with L–L phase separation by depleting one of the solutes, which changes the composition of the remaining liquid phase, delaying (if not preventing) L–L demixing.

For BHJ blends whose phase separation is dominated by L–L demixing, the main role of solvent additives is to prevent or reduce the extent of this demixing. This can be accomplished by inducing earlier aggregation in solution to favor S–L demixing instead, and/or reducing the quench depth into the unstable two-phase region by slowing the solvent evaporation rate. Notably, additives that are poorer solvents than the deposition solvent operate using both mechanisms; however, additives that are good solvents for both donor and acceptor work mainly via the latter mechanism. Moreover, additives can act as compatibi-lizers by reducing the repulsive forces between the donor and acceptor materials and thus hindering large-scale L–L demixing.

In BHJ blends where phase separation is dominated by S–L demixing, phase separation is limited by the nucleation and growth of the separated solid phase. The main role of addi-tives here is to help overcome barriers to crystallization and prolong the crystallization process by slowing the drying of the film. Again, additives of both types (poor solvents and compati-bilizers) were found to improve phase separation in BHJ sys-tems dominated by S–L demixing, although they act through different mechanisms; additives that act as a poor solvent induce early nucleation and extend the crystal growth period by increasing the diffusion of the solutes in the slow-drying film, while compatibilizers largely influence the crystal growth.

The mechanisms by which solvent additives nucleate and pro-mote crystal growth should be distinguished from those of solid nucleating agents. The latter manipulates crystallization pri-marily by reducing the nucleation barrier through heterogeneous

Adv. Mater. 2018, 30, 1707114

Figure 11. a) Schematic representing ternary phase diagram of system composed of solvent (S), acceptor (A), and donor (D) with different donor–acceptor interaction parameter (χDA). Dashed and solid curves represent the binodal and spinodal compositions, respectively. The homogenous mix-ture is thermodynamically stable in the one-phase region, metastable in the region between binodal and spinodal curves, and unstable in the two-phase region under the spinodal curve. The dashed black line represents the change in the blend composition during solvent evaporation; the line corresponds to an arbitrary mixture composition with donor: acceptor (1:2). The onset of L–L phase separation occurs upon crossing from the one phase to the two-phase regions by spinodal decomposition. The onset of S–L phase separation occurs upon reaching solubility limit of the donor or the acceptor in the solvent. b) Schematic representation of the primary phase separation processes from an initially homogenous liquid phase.



nucleation at the agent’s surface, without affecting the diffusion of the solutes. On the other hand, solvent additives that are poor solvents for any of the BHJ components promote nucleation by reducing the overall solvent quality, forcing early supersatura-tion, and contribute to the diffusion of the solutes during the extended drying period. While nucleation agents have been used to manipulate the initial nucleation density in some BHJ sys-tems, either thermal annealing or solvent additives were used to further enhance crystal growth.[100,101] Thus, reducing the nucle-ation barrier alone may not be effective in obtaining an optimal morphology if the growth rate is insufficient; a well-chosen sol-vent additive can offer the significant advantage of manipulating both nucleation and growth mechanisms simultaneously.

BHJ films can be vitrified into an undesirable highly inter-mixed solid solution of donor and acceptor components, with little to no phase separation.[102] For blends that exhibit high glass transition temperatures, kinetic factors lead to this behavior; thermodynamic factors dominate when interactions between highly compatible BHJ components favor the forma-tion of a mixed phase rather than separate domains.[102] For kinetically driven vitrification, solvent additives can reduce the glass transition of the pure and mixed phases within the BHJ, facilitating crystallization and phase separation.[103] In the case of thermodynamic vitrification, additives can reduce interaction between the donor and acceptor and thus minimize the degree of thermodynamic mixing, especially when chosen to selec-tively solubilize one of the BHJ components.

Understanding the thermodynamic and kinetic mechanisms that determine the BHJ morphology (particularly from complex multicomponent mixtures) is of importance to the future devel-opment of OPV materials and deposition methods. Our under-standing of how solvent additives influence these mechanisms in solution-processed BHJ blends is informed by a substantial body of ex situ and in situ morphological studies. Microstructural anal-ysis on static dry films after deposition can resolve the final mor-phology of the film, hinting at the complex impacts of solvent additives on film formation. However, in situ studies more reli-ably probe the progress of film formation and allow us to observe aggregation, crystallization and phase separation processes in real time, including the presence of metastable phases en route.

At present, the majority of solution-processed BHJ OPV active layers are deposited via spin casting, and thus in situ morphology studies have largely probed BHJ evolution during spin casting.[99,104–106] While convenient at lab scale, this tech-nique is not likely to be scalable to large area modules.[107] Given the sensitivity of the BHJ morphology to film formation conditions, it is currently unclear how spin-casting conditions to optimized the BHJ morphology will translate into other deposition methods a priori. However, awareness of meniscus-guided coating (such as blade coating, solution shearing, or slot-dye coating) and other industrial printing methods (like roll-to-roll, inkjet printing, and spray coating) is growing within the solution-processed organic semiconductor community.[5]

3.2. Influences on Liquid–Liquid Demixing

The formation of oversized fullerene domains, often in droplet shapes, is attributed to L–L phase separation for a variety of

BHJ blends, such as PDPP5T, PTB7, PBDTTPD, PDPPTPT, PDTG-TPD, and PBnDT-FTAZ blends with fullerene.[108–113] Processing with an appropriate solvent additive can effectively prohibit the formation of these oversized fullerene domains to obtain the desired morphology for such systems.

For example, in PDPP5T:PC71BM blends, fullerene-rich domains over a hundred nanometers in length were formed when spun cast from pure CF but casting with 5% v/v oDCB prevented the formation of these oversized domains, as shown in the TEM images in Figure 12a.[108] In situ laser reflectom-etry was used to probe film formation to reveal the func-tion of oDCB in this blend. Microscale phase separation and thickness evolution during spin coating of blended films was determined by off-specular scattering and specular interfer-ence, respectively, as shown in Figure 12b.[108] The scattering intensity in the out-of-specular direction—driven by the refrac-tive index contrast between the emerging phases, the PC71BM-rich droplets and the PDPP5T-rich solution—was used to probe the onset of L–L phase separation; the onset of L–L demixing is marked by blue dashed line for the blend film cast directly from CF. For the pure CF case, the deposition solvent com-pletely evaporates within 0.8 s; L–L phase separation started 0.6 s after the start of spin coating, corresponding to a solute concentration of 20 volume percent. In the 5% v/v oDCB case, on the other hand, no strong off-specular scattering signal was observed. Instead, a small scattering step was detected, marked by red dashed line in Figure 12b, and attributed to scattering from small inhomogeneities caused by polymer aggregation. Moreover, the film drying was prolonged to 6 s and the thin-ning behavior appeared to possess two distinct steps: one domi-nated by CF evaporation (up to 0.7 s) and the other dominated by oDCB evaporation (ending at 6 s). During solvent evapora-tion, the solvent mixture becomes concentrated with the higher boiling point solvent oDCB, which is a poor solvent for the polymer PDPP5T and thus induces polymer aggregation.

In a separate experiment using in situ absorbance meas-urements, oDCB was found to force the polymer aggregation during spin casting.[108] The onset of polymer aggregation was found to occur at lower solute concentrations when more oDCB is used in the initial formulation, as shown in Figure 12c. At a certain oDCB concentration (above 3% v/v), polymer aggrega-tion occurred prior to reaching the onset concentration of L–L demixing (≈20% solute), and thus prevented the overcoarsening of PC71BM domains. It is worth noting that inducing polymer pre-aggregation alone by allowing a blend solution in CF to stand for several days did not prevent the formation of oversized PC71BM domains, as films cast from this still exhibited large-scale phase separation. van Franeker et al. hypothesized that using a poor solvent as an additive could induce polymer aggre-gation in a supersaturated solution and lead to the formation polymer fibrillar networks that gelate the solution and increase its viscosity, thus preventing large scale L–L phase separation.[108]

In another BHJ system based on pDPP:PC71BM blends, the effects of solvent additives were investigated[114] using grazing incidence wide-angle X-ray scattering (GIWAXS) and grazing incidence small-angle X-ray scattering (GISAXS).[115] Similar to the previous example, a cosolvent mixture of 4:1 CF:oDCB was found to be effective in attaining a more finely mixed mor-phology.[114] GIWAXS on dry films prepared from pure CF and

Adv. Mater. 2018, 30, 1707114



cosolvent CF:oDCB revealed no change in polymer lamellar stacking spacing (i.e., identical d-spacing) but had slightly larger polymer coherence lengths for the pure CF sample, see Figure 13a. GISAXS, which is sensitive to the contrast in elec-tron-density between pDPP and PC71BM, revealed significant changes in the domain size (see Figure 13b); samples prepared from pure CF had much higher intensity in the low q-range with a very weak shoulder at ≈0.0048 Å−1, while samples prepared from cosolvents exhibited a scattering shoulder at relatively high q-range (≈0.015–0.02 Å−1), indicative of domains with a much lower characteristic length (≈200 Å). Transmission resonant soft X-ray scattering can access a lower q-range (and thus larger structures) than GISAXS, and in the no additive films, revealed the appearance of an additional peak at ≈0.0023 Å−1 (corre-sponding to characteristic length scale of ≈270 nm), besides the weak shoulder at ≈0.0048 Å−1 (see Figure 13c).

Probing the morphology evolution of pDPP:PC71BM using in situ GIWAXS and GISAXS reveals four distinct stages of film formation in the presence of cosolvent, as shown in Figure 13d and represented schematically in Figure 13e.[114] While these in situ experiments were performed on drop-cast films, Liu et al. claim that the solvent additive has a similar effect during spin casting, which is supported by comparable ex situ GIWAXS and GISAXS profiles for drop-cast and spin-cast films probed after deposition.[114] During the initial stage

of solvent evaporation, the rapid increase in oDCB concen-tration (due to CF evaporation) promotes the formation of donor crystallites, as revealed by a steady increase in scattering intensity for the (100) peak of pDPP in Stage I. In the second stage, the increasingly poor solvent quality results in more pDPP crystallization, forming donor fibrillar networks that prevent further large-scale phase separation. Simultaneously, a rapid increase in the characteristic length scale is observed by GISAXS, indicating the onset of phase separation. By the end of this stage, the overall microstructure reaches its maximum characteristic domain size, established by the polymer fibrillar networks. In the third stage, the remaining pDPP crystallizes and PC71BM segregates between the spaces in the fibrillar net-work, with no observable increase in phase separation. In the final stage, solvents evaporate completely and the microstruc-ture evolution reaches a steady state. The drying stages in this study are exaggerated by the high concentration of oDCB addi-tive (20% v/v), but are anticipated to exist at lower solvent addi-tive loadings.[114]

3.3. Influences on Solid–Liquid Demixing

More commonly, solvent additives are used to improve solid-state order and promote coarser BHJ phase separation with

Adv. Mater. 2018, 30, 1707114

Figure 12. a) Bright-field TEM images of PDPP5T:PC71BM films spin coated from CF:oDCB solvent mixture with different oDCB vol% as indicated on the images (scale bar = 600 nm). b) Thickness evolution calculated from specular interference (on top of the graph) and off-specular light scattering (on bottom of the graph) during spin coating of PDPP5T:PC71BM from pure CF (blue) and CF with 5 vol% oDCB (red). c) Schematic represents the role of solvent additive oDCB in preventing the liquid–liquid phase separation by inducing early polymer aggregation prior reaching onset composition of L–L demixing; schematics on the left of the graph c) represent the film formation at the absence of additives where L–L phase separation occurs early in solution (≈80% solvent content) leading to large-scale phase separation; schematics on the right of the graph c) represent the film formation at the presence of solvent additives where the additives (3% or more) induce earlier polymer aggregation before reaching the onset composition of L–L phase separation; the graph in the middle represents the L–L phase separation and polymer aggregation processes as a function of the additive content. Adapted with permission.[108] Copyright 2015, Nature Publishing Group.



more donor-rich domains. For example, in P3HT:PC61BM, the solidification pathway is dominated by S–L phase separation when the blends are cast from single deposition solvents such as CB, oDCB, and TCB.[116] However, the degree of order and phase separation in the blend films is often limited by the fast drying kinetics and vitrification of P3HT by the acceptor. While slowing the drying kinetics with a higher boiling point solvent can enhance P3HT order and phase separation, an additional thermal annealing step is usually needed to further improve the BHJ morphology.[116] Phase separation, in this case, is induced by crystallization; PC61BM is forced out of disordered polymer domains as the P3HT crystallizes, leading to both increased in the order in P3HT-rich domains and the formation of PC61BM-rich domains.[116]

Solvent additives can trigger the S–L demixing mechanism by overcoming barriers to nucleation and growth, in addition to manipulating the kinetics of film drying. Richter et al. inves-tigated the role of additives ODT and 1CN in manipulating the BHJ morphology of P3HT:PC61BM.[117] PC61BM significantly hinders P3HT crystallization, evidenced by reduced relative crystallinity of P3HT in blend films compared to neat films in Figure 14a. Interestingly, ODT (a poor solvent for P3HT) and 1CN (a good solvent for P3HT) operate by different mecha-nisms but create similar final BHJ morphologies.[117] While both ODT and 1CN improve overall P3HT crystallinity, pro-cessing with 1CN increased the mosaic-character of P3HT crystalline domains revealed by pole figures of (100) peak (Figure 14b) indicated by its broader scattering profile.

Adv. Mater. 2018, 30, 1707114

Figure 13. Scattering intensity versus q plots obtained from a) GIWAXS, b) GISAXS, and c) R-SOXS of pDPP:PC71BM blends prepared from pure CF and with cosolvent CF:oDCB; solid and dash-dot lines in (a) and (b) correspond to as cast and annealed thin films, respectively. d) In situ GIWAXS and GISAXS results obtained from drop-casting pDPP:PC71BM from CF:oDCB (4:1); (100) peak scattering intensity and (100) d-spacing of pDPP obtained from GIWAXS are represented as blue and red open squares, respectively; domain size of scattering particles obtained from Guinier analysis of GISAXS is represented as black squares; solvent weight change is represented as green open squares. e) Schematic of the film formation process during the four stages marked in (d). Adapted with permission.[114] Copyright 2012, Wiley-VCH Verlag GmbH & Co. KGaA.



The microstructure evolution during film formation, sum-marized schematically in Figure 14e, was probed using a variety of techniques, including: p-polarized near-Brewster transmission spectroscopy to probe the evolution of absorbance spectra, indicative of the progression of photophysical aggre-gation; GIWAXS to probe P3HT crystallization behavior; and GISAXS to probe the formation of domains.[117] In the absence of solvent additives, photophysical aggregation and crystalliza-tion develops at the end of film drying and its progress ends abruptly, consistent with fast CB evaporation—see the top

schematic of Figure 14e. P3HT in blend films cast from pure CB solution tends to form highly oriented edge-on crystallites, indicated by the pole figure in Figure 14b.

1CN induces crystallization of P3HT slightly earlier than the deposition solvent CB, which is attributed to the solvent addi-tive reducing the polymer’s nucleation barrier; this is likely due to the lower solubility of P3HT in 1CN than in CB.[117] Notably, the crystallization period is prolonged significantly by the slower evaporation of 1CN. As with deposition from CB, initial P3HT crystallization is predominantly oriented edge-on; however,

Adv. Mater. 2018, 30, 1707114

Figure 14. a) Relative crystallinity of P3HT calculated for neat and blend samples coated from pure CB and with additives. b) Integrated pole figure (scattering intensity vs orientation angle) of samples in (a). c,d) In situ GIWAXS and GISAXS results of P3HT:PC61BM blend during casting from CB with 1CN in (c) and CB with ODT in (d); evolution of integrated pole figures of (100) peak of P3HT (red); evolution of TSI (gray); computed TSI with two-phase and three-phase models (green). e) Schematics represent film formation when casting P3HT:PC61BM blend from pure CB (top of the graph), CB with 1CN (middle of the graph), and CB with ODT (bottom of the graph). Adapted with permission.[117] Copyright 2015, Wiley-VCH Verlag GmbH & Co. KGaA.



further crystallization driven by 1CN evaporation results in a mix of orientations, which explains the increased mosaic-char-acter observed ex situ for films cast with 1CN (Figure 14b). This increase in number of small crystalline domains is attrib-uted to more bulk-dominated crystallization compared to the interface-dominated pure CB casting condition.[117] The evolu-tion of domain formation was explained by a two-phase model, obtained by fitting the total scattering invariant (TSI) from GISAXS results, as shown in Figure 14c. The two emerging phases during 1CN evaporation were assigned to a crystalline P3HT phase and a mixture of 1CN, P3HT, and PC61BM such that the phase separation dominated by P3HT crystallization, as portrayed in the middle schematic of Figure 14e.[117] The additive 1CN, being good solvent for both P3HT and PC61BM, works by solvating this mixed phase, allowing P3HT in this phase to crystallize and extrude PC61BM.

When P3HT:PC61BM is processed with ODT, the observed film formation behavior is similar to those processed with 1CN, but with slightly earlier donor aggregation than in CB and pro-longed crystallization.[117] However, a three-phase model better explains the evolution of domain formation in ODT-processed films rather than the two-phase model used for 1CN. Here, the emerging phases are attributed to crystalline P3HT, a mixed phase of P3HT:PC61BM, and dissolved PC61BM in ODT, as shown in Figure 14d,e. Thus, ODT also functions by selectively dissolving PC61BM, which depletes the fullerene in the mixed

phase and facilitates P3HT crystallization—a distinct mecha-nism that was not observed for 1CN.[117]

Overall, for BHJ systems that are dominated by S–L phase separation, the primary function of solvent additives is to pro-mote the nucleation and growth of the emerging separated solid phase even though the final microstructure may vary in terms of the nucleation density and extent of crystal growth.[82,117] Spe-cific additives may affect whether the crystallization is induced dominantly in the bulk solution or at interfaces (e.g., solution/air or substrate/air interfaces), leading to a change in the pre-ferred molecular orientations.[82]

3.4. Inducing Phase Separation in Intimately Mixed Polymer:Fullerene Blend Morphologies

The microstructural evolution of the PCPDTBT:PC71BM film was investigated using in situ GIWAXS just after spin coating from pure CB and from CB with 3% v/v ODT.[118,119] As-cast films exhibit poor polymer ordering indicated by an isotropic amorphous scattering pattern that did not change over time. In additive-processed films, as shown in Figure 15a, the formation of PCPDTBT crystallites was observed immediately after spin coating, and crystallization continued for more than an hour, as revealed by an increase in the total scattering intensity of polymer lamellar stacking peak. The observed polymer crystallization

Adv. Mater. 2018, 30, 1707114

Figure 15. a,b) In situ GIWAXS results obtained after spin coating of PCPDTBT:PC71BM from CB with 3% ODT; a) time evolution of total scattered intensities of the lamellar stacking peak of PCPDTBT and the amorphous region (red); b) time evolution of the peak position of the (100) peak (blue) and the (100)’ peak (red). Adapted with permission.[118] Copyright 2012, American Chemical Society. c,d) TEM images of PCPDTBT:PC71BM films cast from pure CB and CB with 2.5% DIO, respectively. e) Schematic represents the morphology of PCPDTBT:PC71BM films with and without additives for samples in (c) and (d). Adapted with permission.[119] Copyright 2012, American Chemical Society.



earlier during film formation was attributed to a reduced poly mer nucleation barrier brought on by the presence of ODT, a poor solvent for PCPDTBT. Interestingly, two polymorphs of the polymer were observed for the additive-processed sample, with distinct lamellar stacking peaks; one was mainly oriented out-of-plane near q ≈ 5 nm−1, labeled (100), and the other oriented in-plane near q ≈ 5.5 nm−1, labeled (100)′. The evolution of peak position for (100) and (100)′ is plotted in Figure 15b, and shows that the alkyl spacing within polymer crystallites oriented out-of-plane condense over time, indicated by the shift of the (100) peak position towards higher q values (i.e., lower d-spacing). The time-dependent behavior of this shift appears to track with the evapo-ration of the additive DIO, demonstrating that the mechanism of this solvent additive can be used to promote S–L demixing even in blends with relatively amorphous donor polymers.

Another study of the PCPDTBT:PC71BM system used a variety of solvent additives, including DIO, ODT, and 1,8-dichlo-rooctane.[119] As before, the additive-free PCPDTBT:PC71BM film appears uniform with a finely mixed morphology, as shown in Figure 15c. The formation of intimately mixed morphology for additive-free samples was supported by small-angle neutron scattering (SANS) and R-SoXS measurements, where SANS and R-SoXS scattering profiles appeared featureless, indicative of a scattering from a uniform mixture of PCPDTBT and PC71BM. Formation of polymer crystalline fibrils and improved phase separation were observed in blend films processed with any of the three additives, as shown in Figure 15d. All three improved device performance with the highest improvement obtained using DIO. Gu et al. investigated the structural evolution of drop-cast blends from pure CB and CB with 2.5% DIO using in situ GIWAXS.[119] The pure CB-processed sample appears amorphous; only PC71BM scattering features are observable. Adding DIO induces crystallization of PCPDTBT and the crystallization process continues within the blend film for an hour after casting. A schematic representing the morphology of the BHJ with and without additives is presented in Figure 15e.

An additional study investigated the impact of DIO on PCPDTBT:PC71BM but from a different deposition solvent, 1,2,4-trichlorobenzene. Peet et al. monitored the progression of photophysical aggregation during solvent evaporation using UV–vis absorption measurements.[47] When the additive was used, two distinct polymer aggregation steps were observed; one was driven by the depletion of the host solvent TCB (pro-moting S–L demixing) and the other driven by the slow evapo-ration of the higher boiling point solvent DIO (extending S–L demixing). Moreover, DIO prolonged the polymer aggregation process significantly, leading to higher degree of polymer order and chain planarization for the additive-processed sample as compared to the sample coated from pure TCB.[47]

3.5. Studies of Small Molecule:Fullerene Blends

While both types of additives, poor solvents and compatibilizers, demonstrate enhancement with P3HT:PC61BM and other BHJ systems, this is not the case for all BHJ blends. For instance, for films of il(TT)2 with PC71BM, only poor solvents for the donor brought the film close to an optimum BHJ morphology.[82] All of the additives studied, including poor and good solvents,

formed crystalline fibrils of il(TT)2 but the size and distribution of the fibrils varied significantly; good solvents resulted in over-growth of crystalline domains, while poor solvents increased the nucleation density leading to the formation of smaller and interconnected fibrils.[82] This deviation from trends observed for polymer-based active layers has been be attributed to small molecules’ comparatively greater propensity for crystallization such that S–L demixing supersedes L–L demixing processes.

Studies on small-molecule BHJ systems have also revealed additional correlations between the crystallization pathway and dynamics of the donor molecules and the need for solvent additives.[90,92,93] In particular, deposition from a single solvent is often insufficient to form an ideal BHJ morphology due to kinetic barriers to crystallization.[120] An in situ GIWAXS study on p-DTS(FBTTh2)2:PC71BM system shows the formation of a weakly ordered metastable polymorph of p-DTS(FBTTh2)2 when casting from pure CB. This polymorph exhibits broad and weak scattering with no indication of long-range order, as shown in Figure 16a.[90] Moreover, the dried blend films exhibit an intimately mixed morphology with poor phase sep-aration, as shown in Figure 16c. The additive DIO promotes the transformation of the metastable phase to a highly stable crystalline phase and enhances phase separation with the appearance of interconnected wire-like crystalline structures of p-DTS(FBTTh2)2, as shown in Figure 16b,d.[90,92]

The metastable phase of p-DTS(FBTTh2)2 was identified as a transient liquid crystalline phase in a subsequent work that extended the study to a variety of structurally similar donor molecules—p-SIDT(FBTTh2)2, p-DTS(FBTTh2)2, and X2.[93] This investigation correlated the ease of crystallization of these donor molecules to their need for solvent additive processing.[93] In situ GIWAXS and flash differential scanning calorimetry (flash-DSC) were used to investigate the ease of crystallization from solution and melt, as summarized in Figure 17.[58,93,217,218] Donor molecules that are easily vitrified by rapid film forma-tion, either by being trapped in transient liquid crystalline phases (such as p-DTS(FBTTh2)2) or amorphous phases (such as p-SIDT(FBTTh2)2), require solvent additives to overcome barriers to crystallization. On the other hand, donor materials with strong self-assembly tendencies (i.e., with a low activation barrier for crystallization like X2) readily crystallize and phase separate, without the need for additives.[93] These features may be used to rationalize why some materials are less responsive to solvent additive processing.

In another in situ GIWAXS study, the solid additive polysty-rene (PS) was used concurrently with the solvent additive DIO to synergistically affect the BHJ morphology and improve device performance.[92] Inclusion of the PS was found to induce greater crystallization of p-DTS(FBTTh2)2 within the film (see Figure 16e) as compared to samples cast from pure CB. In addition, crystal-lization of p-DTS(FBTTh2)2 occurred primarily at the free surface (liquid/air interface), while the bulk was kinetically trapped in the metastable phase. This crystallization behavior was attributed to favorable nucleation on the surface of the PS-rich domains and CB retention within them; the retained solvent subsequently diffuses to the surface, inducing crystallization along the diffu-sion path, evidenced by increased donor fibril formation. While PS improved the overall BHJ morphology when compared to the additive-free condition, the obtained final morphology was not

Adv. Mater. 2018, 30, 1707114



yet optimal for OPV performance. Combining this effect with the extended drying time of DIO was found to promote further p-DTS(FBTTh2)2 crystallization (see Figure 16e) and obtain better device performance than either additive alone.[92]

3.6. More Complex Additive Combinations

Solvent additive processing has dominated the BHJ OPV field precisely because it provides multiple readily accessible

experimental handles for understanding and manipulating the existing film formation processes during deposition. However, solvent additives are part of a larger toolkit for manipulating film formation, complementing blend solution optimization and postprocessing techniques. For example, changing the deposition solvent or using a cosolvent also drastically impact the resulting BHJ morphology by manipulating L–L or S–L demixing.[81] Other techniques can be used to manipulate the drying kinetics during or after casting, such as solvent vapor annealing,[37] controlled vapor chambers[37,121] or inverted

Adv. Mater. 2018, 30, 1707114

Figure 17. Comparison of the ability of crystallization, crystallization behavior, and processing routes between p-SIDT(FBTTh2)2, p-DTS(FBTTh2)2, and X2. The compounds are ranked by ease of crystallization from lowest (p-SIDT(FBTTh2)2) to highest (X2). Quenching refers to the cooling speed from the melt that does not exhibit crystallization by flash DSC, while vitrification refers to the heating speed from the quenched melt without cold crystallization. Data for (a)–(d) are taken from refs. [58,217,218] and ref. [93], respectively. Reproduced with permission.[93] Copyright 2015, Wiley-VCH Verlag GmbH & Co. KGaA.

Figure 16. a,b) 2D-GIWAXS plots of integrated scattering intensity versus q versus time during spin coating and thereafter of p-DTS(FBTTh2)2:PC71BM blends from a) CB and a,b) CB with DIO. Reproduced with permission.[90] Copyright 2013, Wiley-VCH Verlag GmbH & Co. KGaA. c,d) Energy-filtered transmission electron microscopy images of p-DTS(FBTTh2)2:PC71BM samples prepared from c) CB and d) CB with DIO. e) GIWAXS-integrated scat-tering intensity versus q for p-DTS(FBTTh2)2:PC71BM samples cast from pure CB, CB/DIO, CB/PS, and CB/PS/DIO at selected times during spin casting and thereafter. Reproduced with permission.[92] Copyright 2015, Wiley-VCH Verlag GmbH & Co. KGaA.



drying.[122] Thermal annealing can be readily combined with solvent additives but is typically used to remove residual addi-tive rather than for further morphological control. However, even gentle annealing can induce additional morphology evolu-tion, and brief high temperature annealing directly after casting has been shown to form donor fibrils as the solvent additive rapidly evaporates toward the top surface of the film.[123]

Each of the methods listed above can conceivably be used, alone or combined, to manipulate the BHJ and thus enhance performance.[124] However, not all methods readily approach the optimum morphology for a particular donor:acceptor blend because they operate via different mechanisms. Thus, combining multiple strategies can target and rectify specific morphological deficiencies, often with synergistic effects on efficiency. There are numerous examples of binary additives providing fine control over the phase separation of BHJ blends, e.g., Hou and co-workers used 20% v/v CF cosolvent and 5% v/v DIO in oDCB to create finer phase separation and purer domains in PDPP3T:PC71BM.[125] Wang et al. demonstrated two solvent additives, DIO and 1CN, can be used synergistically for PDPPFTF:PC71BM; as-cast from CB, the acceptor aggregates into large domains, resulting in low FF of 0.33 and PCE of 0.9%. Processing with either 3% v/v DIO or 1CN results small iso-lated acceptor aggregates, improving PCE to 4.4%. Using both additives yield PC71BM domains that are simultaneously fine and interconnected, optimizing the PCE to 4.7% and improving the diode character of the device.[126] Pairing additives with dif-ferent mechanisms of enhancement is clearly more important than either’s chemical identity; for example, Wan et al. extended this strategy to blends of PTB7-Th (and two other polymers) with PC71BM that over phase separate when cast from CB, using NMP to increase miscibility and DIO to boost donor crys-tallinity; for PTB7-Th:PC71BM, the PCE is 8.1% as-cast, 8.9% with NMP, 9.2% with DIO and an impressive 10.4% when both additives are used. Impedance spectroscopy showed that films optimized with both additives have significantly better charge transport and balanced electron and hole mobilities, indica-tive of bicontinuous networks and phase purity.[127] Similarly, Jung et al. used the polymeric additive PDMS to nucleate donor fibrils while DIO suppressed acceptor aggregation in PBDTTT-CF:PC71BM films—increasing PCEs to 7.3% compared to 2.6% as-cast and 6.6% with DIO alone.[128] Interestingly, the use of multiple additives may expand the processing window for optimal performance observed for each additive alone despite the sensitivity of the BHJ morphology to additive concentration. Sharenko et al. used a commercial nucleating agent, DMDBS, to seed crystal growth homogenously throughout the film to limit donor overgrowth in the presence of excess DIO.[100] Pro-cessing p-DTS(FBTTh2)2:PC71BM with 1.5% DIO, greater than the optimum 0.4%, decreases PCEs from 7.0% to 0.7%; such blends with DMDBS only drop to 4.2%.

4. Use of Solvent Additives in Related Electronic Technologies

The morphology of any solution-processable organic semicon-ductor (OSC) could benefit from the use of solvent additives but few OSC fields use additives as extensively as OPVs. Other OSC

fields often have fewer constraints imposed by morphological complexity when compared to BHJ OPVs, either due to lower concentration requirements for single components (common in organic field effect transistors), competing parameters that lessen performance dependence on morphology (as in organic thermoelectrics) or sequential deposition steps (common in perovskites). As such, understanding the mechanism of sol-vent additive enhancement in these applications are smaller research fields and often draw their inspiration from OPV mor-phological studies.

In this section, we provide selected examples of solvent addi-tives in the solution processing of other types of devices and materials, with dual emphasis on common mechanisms from Section 3 and functions unique to that field.

4.1. Organic Field-Effect Transistors (OFETs)

OFETs are typically three terminal devices wherein current passes through a semiconducting material between a source and a drain electrode, atop a dielectric layer. Typically, the semi-conductor layer consists of a single material, although blend films similar to the BHJ in OPVs have also been studied. A key figure of merit for OFETs is the charge carrier mobility, μ. Most organic semiconductors exhibit some degree of ambipolar transport, modulated by the device architecture.[129–131] Higher mobilities are typically the result of increased intermolecular π–π overlap area, short π–π stacking distances, and long-range order. Thus, the measured mobility strongly depends on the active layer purity and degree of structural order across multiple length scales, as well as on the environmental conditions of the measurement.[132]

Choosing an appropriate deposition solvent is the primary strategy for tuning the OFET morphology, manipulating drying time[133–135] or solubility.[136] Thin-film OFET fabrication imposes fewer solubility requirements than in OPV applications, typi-cally having to accommodate only one solute under dilute conditions (<10 mg mL−1). Much research effort has focused on aligning crystallites within the OSC active layer,[137,141] including directing the evaporation of the deposition solvent. These range from simple methods, such as off-center spin coating[138] and drying under a tunnel-like structure,[139,140] to complex flow chambers.[37] Nonetheless, solvent additives are more readily applied and scaled than these methods, and so the use of additives remains common in the OFET field.

A wide variety of solvent additives have been used on OFETs, and their effects can be linked to the same mechanistic princi-ples outlined in Section 3. For example, Ichikawa et al. found that trifluoromethylbenzene (TFMB) is an effective additive for increasing the crystallinity and thus mobility of P3HT thin films. However, enhancement only occurs for deposition sol-vents with lower boiling points than TFMB, such as carbon tet-rachloride (CCl4), CF, and THF. This effect is not due to doping interactions, as similar improvement was observed for methyl-cyclohexane, a poor solvent with similar vapor pressure to TFMB but different electronic structure.[142] Later work found that additives with good solubility for the active layer, such as trichlorobenzene, enhance performance, suggesting that low vapor pressure is also a factor.[54] Taken together, this suggests

Adv. Mater. 2018, 30, 1707114



that poor solubility and extended drying times are important selection rules for solvent additives in OFET applications. Given the lowered requirements, computational studies using Hanson solubility parameters have been particularly useful for identifying “dissimilar solvents” for established OFET mate-rials, such as TIPS pentacene.[143]

It makes sense that additives used in OPVs are often suc-cessful in OFETs, given their similar chemical structures. In a detailed morphological study, Liu et al. reported that similar concentrations of DIO that lead to enhanced mobility in OPV blend films yield optimum transport in donor-only OFETs for several DPP-based polymers, due to the formation of donor fibril structures.[144] An et al. similarly found that 1CN seeded more crystallites of a DPP-based polymer than in as-cast films, and their mobility could be increased by subsequent thermal annealing.[145] Park et al. found that small variations in DIO or 1CN concentration drastically changed the morphology of isoindigo-based small molecule thin films, which correlated well to the observed changes in mobility. Minute amounts of DIO (0.2% in CB) increased the number and size of crystal-line grains for all four molecules studied, increasing mobili-ties. Similar concentrations of 1CN resulted in smaller grains and smoother morphology, useful for some derivatives.[146] Enhanced mobility in OFET active layers can result from bal-ancing overall crystallinity (promoted by poor solvents) with few grain boundaries (promoted by good solvents), which depend both on the interplay between the OSC material and the deposition solvent.[132]

While charge transport occurs throughout the active layer film in OPV applications, transport in OFETs is confined to a thin layer adjacent to the gate, typically at the semiconductor/dielectric interface. Thus, the ability of solvent additives to induce vertical phase separation can be exploited in OFETs to tune the dominant transport character in BHJ films. For example, Park et al. demonstrated this using DIO and 1CN additives with a p-type small molecule DTTDPP and n-type PC71BM, which form ambipolar BHJ OFETs. Processing with DIO causes the p-type molecule to aggregate in the center of the active layer, as shown in Figure 18, facilitating electron

transport through the fullerene phases to create an n-type tran-sistor. Conversely, the evaporation of 1CN causes significant vertical phase segregation such that the donor molecules aggre-gate primarily on the dielectric layer, increasing hole mobility to create a p-type transistor.[147]

Finally, residual high boiling point deposition solvents and solvent additives may increase the long-term efficiency of OFET transport. Nikolka et al. showed that various residual solvent additives can fill voids in the active layer of three polymers with significantly different chemical structures. The effect is similar to that of a nonvolatile solid additive, like F4TCNQ, but diminishes as the additive evaporates (typically over a month). These voids would otherwise fill with air or water molecules that lead to rapid performance decreases when oxidized.[148] This enhancement mechanism is unique to OFETs, as residual solvent in OPV active layers has been shown to contribute to photodegradation.[149]

4.2. Organic Thermoelectrics (OTEs)

In thermoelectric materials, charge carriers move within the film in response to a temperature gradient, creating a poten-tial difference across the material, known as the Seebeck effect. This potential gradient can be used to generate power from ambient heat, or reversed for converting electricity directly to heating or cooling applications. Thus, an OTE material should exhibit higher electrical conductivity than thermal conductivity to maintain the potential and thermal difference simultane-ously. For near-ambient temperature applications, the thermal insulating properties of organic semiconductors are particularly promising, provided their charge mobilities are sufficiently high. Recent progress on the design and processing of OTEs is summarized in various reviews.[150–152]

Progress in OTEs draws heavily from advances in high-carrier mobility structures and strategies from OFETs and conductive layers in OPVs, including the use of solvent addi-tives to manipulate morphology. Intrinsic organic semicon-ductors also used for OPVs would conceivably benefit from

Adv. Mater. 2018, 30, 1707114

Figure 18. Schematic illustrations of proposed distribution of p-type (green) and n-type (purple) components in BHJ OFET devices. Adapted with permission.[147] Copyright 2013, American Chemical Society.



Adv. Mater. 2018, 30, 1707114

the additional crystallinity granted by solvent additives, like DIO.[152] Improving the crystallinity and alignment of OTE materials improves electronic conduction but often simultane-ously decreased thermopower; a mix of ordered and disordered domains often leads to optimal thermoelectric performance.[44] Because of this diametrically opposed relationship and without the need to absorb visible light, the majority of OTE materials differ significantly from structures used for OPV active layers. For example, the most widely studied OTE is the water-process-able dispersion PEDOT:PSS, a doped polymer composite with conductive and insulating components. PEDOT:PSS is com-monly used as a visibly transparent hole-transport/electron-blocking layer in conventional OPV devices.

In such composite materials, solvent additives may act as dopants, compatibilizers, or morphological control agents. Phase separation of the PSS and PEDOT components in PEDOT:PSS films impacts the degree of doping as well, which has a stronger impact on mobility than just morpho-logical ordering.[153] Both DMSO and ethylene glycol have been used as deposition solvents[154] and as additives[155,156] to enhance conductivity in this manner. The same mechanism was reported to enhance another EDOT derivative, PEDOT:Tos, by using a variety of solvent additives.[157] Eutectic mixtures of solvents (e.g., ethylene glycol and amine, ammonia and water) with lower vapor pressure than either solvent alone have been used to greatly increase evaporation time and thus regulate morphology.[158,159]

4.3. Perovskite Solar Cells

Organic–inorganic hybrid perovskite crystals have seen recent success as solar cells materials, with PCEs increasing from 3.8% in 2009 to over 22% in 2016.[160] This rapid advance is largely due to optimization of the perovskite film morphology by manipulating crystal nucleation and growth during solution processing.[161] Like in BHJ OPVs, the surface morphology and the quality and size of the perovskite crystallites strongly impact performance. However, fabrication methods commonly involve two or more deposition steps, and various solid and solvent additives can be added sequentially to manipulate the crystalli-zation kinetics. Thermal annealing and solvent vapor annealing treatments are also used. Additive engineering[162] and optimi-zation of the various solvent deposition processes[160,163] are active areas of research within this field.

Solvent additives for morphology manipulation in per-ovskites can be generally categorized into two methods of enhancement: additive-induced crystallization, where a poor solvent nucleates the perovskite lattice, or solvent mediated crystallization, where a high-boiling point solvent extends the timeframe for crystalline growth. This is analogous the impact of DIO and 1CN on S–L demixing in OPVs, respectively. Both DIO and 1CN additives are reported to enhance perovskite per-formance; they function as chelation agents for the Pb2+ pre-cursor and remain in the film after deposition to delay crys-tallization. Liang et al. reported that including 1% v/v DIO in the lead halide precursor solution encourages homogenous nucleation and then slows the growth of perovskite crystals, leading to a smoother and continuous active layer.[164] Other

high-boiling point solvents capable of chelating Pb2+ demon-strate similar PCE enhancement—including bidentate ligands like alkyl halides[165] and DMSO,[166] and monodentate ligands like 1CN[167,168] and acetonitrile.[169] Notably, both inducing early nucleation and delaying crystal growth using solvent addi-tives results in more uniform perovskite films than extended thermal annealing, which can cause large crystals to peel away from the substrate surface. Selective solvation can be used to achieve both effects simultaneously: for example, Mo et al. included a small amount of a polar solvent (DMF or gamma-butyrolactone, GBL) to solvate the lead halide precursor, allowing the PbI2 to more effectively diffuse throughout the active layer.[170]

The performance and stability of perovskite solar cells is sen-sitive to the presence of water.[171] Surprisingly, a certain degree of humidity during fabrication can reportedly nucleate pero-vskite crystals and improve performance. Indeed, trace mois-ture in organic solvents or the fabrication environment likely served as an inadvertent solvent additive in early perovskite development. While high levels of water during and after fab-rication irreversibly disrupt the perovskite lattice, Gong et al. demonstrated that limiting the water content to 2% v/v in DMF led to an optimized morphology for single-process perovskite solar cells, with PCEs of 16.1% (wet) versus 12.1% (dry).[172]

5. Future of Solvent Additive Processing in Organic Photovoltaics

The impetus behind organic photovoltaics has consistently included consideration of their environmental impact. In search of higher efficiency, the field has diversified chemical structures, deposition processes, and device architectures. Although reliably high PCEs have been achieved with multiple OPV systems,[173–175] silicon, thin film (e.g., CdTe), and more recently perovskite solar cells continue to outperform OPVs in terms of efficiency. Research has started to shift in response to highlight organic semiconductors’ unique and desirable mate-rial properties: color tunability, lightweight, thin, printable, flexibility, conformable, biocompatible, and elements of low toxicity. The current challenges facing solution-processed OPVs concern their long-term device stability, sustainable sourcing and large-scale manufacturing—returning to environment- centric motivations.

Po et al. highlight five primary industrial considerations for the future of BHJ OPVs, three of which are relevant to the processing of the active layer: 1) materials must be stable and soluble enough to form highly concentrated inks that can be used in industrially relevant printing processes; 2) solvents and additives should be readily removable or recoverable, and pref-erably nontoxic and sustainably sourced; and 3) the layer mor-phology should be stable under mild processing conditions.[176] These sentiments are echoed by Kang et al. in “Key Core Tech-nologies for [BHJ OPV] Commercialization”—being module design, flexible transparent electrodes, device efficiency, device stability, and printing technology.[177] Again, three of these chal-lenges relate to the processing of the active layer. While solvent additives can address particular morphological deficiencies, their use in large-scale manufacturing poses new concerns over



Adv. Mater. 2018, 30, 1707114

reproducibility in fabrication, long-term performance stability and environmental toxicity of BHJ OPVs.

In enabling high power conversion efficiencies, additives also underscore how sensitive the BHJ morphology is to fabri-cation conditions. In the majority of cases, the donor:acceptor blend film is quenched into a less desirable organization when processed from a single solvent; high boiling point additives extend the drying process and allow the film to evolve toward thermodynamic equilibrium. However, the BHJ morphology for optimal charge generation and collection for a given donor:acceptor blend does not necessarily coincide with the thermodynamic equilibrium morphology of the system. Thus, some BHJ blend systems are susceptible to overprocessing (e.g., via prolonged thermal annealing or excess solvent addi-tive concentration) and thus lowered OPV performance. These systems would likely benefit from the simultaneous use of multiple morphological agents to widen the processing window.[100] The sensitivity of OSCs to dilute concentrations of solvent additives may also be problematic at scale, as volatile deposition solvents may vary in additive concentration after long-term storage and thus contribute to batch-to-batch vari-ability. This variability extends to the deposition environment as well, as factors such as the rate of residual solvent evapora-tion can affect OPV performance.[178] Furthermore, industrial implementation may require additional solid or solvent addi-tives to modulate the properties of the deposition ink, and fur-ther processing steps to remove residual solvent (like heating or washing).[86,176] BHJ blend systems that are readily manipu-lated using processing additives are more likely to sensitive to these processes.

It is worth pointing out that residual solvent additives can also decrease OPV efficiency in various ways. Chang et al. found that P3HT crystallites formed preferentially at the top surface of P3HT:PC61BM films during the slow evaporation of DIO, leading to poor Ohmic contact.[178] Gently heating the

film after deposition speeds the evaporation without depleting the surface of PC61BM domains. Processing under an inert atmosphere, common in OPV processing, may cause more additive to be retained than under ambient air, as the film can reabsorb solvents from the contained environment.[178] Ye et al. later confirmed that how DIO is released from the active layer impacts the morphology and thus performance for seven dif-ferent polymer blend films processed from oDCB.[149] The researchers found that rapid removal of residual DIO results in higher PCEs, as shown in Figure 19, and suggest two methods to accomplish this: first, placing the blend film under high vacuum, commonly done during electrode evaporation, and second, using a high vapor pressure antisolvent to wash out the DIO, such as methanol or isopropyl alcohol.[149] Thus, the dep-osition and drying environments play significant roles in the OPV performance processed with solvent additives.

Removing residual additives additionally increases long-term device stability, as the presence of halides in the active layer may lead to photodegradation. Tournebize et al. noted that P3HT:PC61BM devices processed with ODT are more sensitive to both light and oxygen than those cast directly from CB.[179] The researchers found that ODT oxidizes when exposed to the intense light of a solar simulator, and that the gaseous photoproducts cannot escape the film once the top electrodes are deposited. This effect was greatly reduced if the active layer was exposed to bright light prior to electrode depo-sition. The authors note that such a light-soaking step could be easily integrated into industrial manufacturing protocols. Tremolet de Villers et al. observed similar effects for PTB7 and PTB7-Th:PC71BM when processed with DIO;[180] the presence of residual DIO in the active layer significantly increased the photodegradation rate of both donor and acceptor components, later observed for other alkyl halides.[181] Removing residual sol-vent additive thus requires special attention given the potential for additional degradation pathways.

Figure 19. Schematic detailing how residual solvent additive is removed from the active layer impacts the BHJ morphology, and thus performance, of polymer:fullerene solar cells. Washing the blend film with an antisolvent, such as methanol, can remove residual solvent additive quickly, replicating exposure to high vacuum. Reproduced with permission.[149] Copyright 2013, American Chemical Society.



Adv. Mater. 2018, 30, 1707114

Ideally, molecular design of the donor:acceptor pair could be tailored such that the blend forms an ideal BHJ morphology when cast from a single deposition solvent; there is no inherent need for additional processing if the readily accessible equilib-rium morphology can provide optimal performance. Indeed, multiple high-performance donor:acceptor pairs have been identified which fit this description, e.g., PBDTSe-TT achieves 8.8% PCE when blended with PC71BM from CB, without the need for DIO unlike its structural derivative, PTB7. Chang et al. attribute the greater as-cast phase separation to selenium’s enhancement of backbone π–π interactions.[182] Zhang et al. changed the deposition solvent for PBDT-TS1:PC71BM from oDCB (6.9% PCE) to 2-methylanisole (9.7%) to directly reach a similar BHJ morphology to oDCB processed with DIO (9.5%).[183] Fluorination increases the crystallinity of the polymer acceptor P(NDI2OD-2FT) such that it drives S–L demixing in blends with PTB7-Th processed from oDCB; DIO and 1CN delay the onset of acceptor crystallization, reducing the as-cast PCE from 6.7% to 1.5% and 4.7%, respectively.[184]

Note that each of these cases resulted from particular interac-tions between the deposition solvent and solutes, with no clear structural pattern—we do not yet have guidelines for designing and selecting additive-insensitive BHJ materials. Encoding this desired self-assembly behavior into the molecular design of an organic semiconductor requires a deeper understanding of structure–property relationships, one that is actively being pursued. Susceptibility to processing additives can be used to assess the film formation tendencies of solution:donor:acceptor blends to identify more deposition-tolerant systems. Suscepti-bility to solvent additives can be a useful metric for assessing the suitability of a material (or blend) for further study and pos-sible industrial implementation, beyond sole consideration of its optimal efficiency.

Common deposition solvents (i.e., CF, CB, and oDCB) and solvent additives (i.e., DIO, and 1CN) are both toxic[185–187] and energy intensive to synthesize;[188,189] the use of “greener”

solvents and additives with lower toxicity and sustainable man-ufacture is preferable.[190] However, environmentally benign solvents and additives are typically nonaromatic and polar in nature, and thus are less likely to be good solvents for current OPV materials. Representative solvents and additives used in OPV processing are classified by their type in Figure 20, ranked by solvent polarity.[191] The side chains and conjugated subunits of both donor and acceptor components will need to be modified to increase solubility in a wider array of media. New additives will also be required to replicate the complex relationships between the BHJ components and deposition solvent.[191] However, number of reports show that nonhalogen-ated and nonaromatic solvents can be used to fabricate OSCs with PCEs greater than 5%.[174,192–194] Also, it is not uncommon for BHJ blends in the emerging green solvent literature to be optimized without solvent additives or thermal annealing, as a greater variety of deposition solvents are tested, e.g., Lee et al. found that PTB7:PC71BM processed from trimethylbenzene performed better films optimized for deposition from CB with DIO.[195]

Until donor and acceptor molecules are designed that self-assemble well from such green solvents, most pairs are likely to form nonoptimized BHJ morphologies as-cast, which solvent additives could readily improve. In green-processed OPVs, new additives have emerged that were not previously tenable: for example, efficiencies of 6.6% were recently reported for PCDTBT:PC70BM processed from carbon disulfide with acetone as a cosolvent.[196] So far, these addi-tives appear to operate via the same mechanisms described in Section 3, but this will need to be borne out by further in situ film formation studies. It is unclear whether they pose similar concerns for long-term stability as halogenated additives like DIO and 1CN.

Finally, polymeric and molecular nonfullerene acceptors (NFAs) have recently emerged that utilize modular donor–acceptor molecular architectures, similar to prominent donor

Figure 20. General categories of solvents used in BHJ solar cell fabrication with representative examples. Median lethal dose (LD50) values (in g kg−1 for rats via oral ingestion, where available) provide a relative measure of toxicity, chosen on the basis of its availability for the listed solvents. Solvent additives are marked with *. Reproduced with permission.[191] Copyright 2017, Elsevier.



Adv. Mater. 2018, 30, 1707114

materials.[197–199] This new class of acceptors offers several advantages over fullerenes such as widely tunable energy levels, greater light absorption, greater solubility in a wider range of solvents.[200] Solvent additives often play a prominent role in optimizing donor:NFA morphologies, as NFAs can be prone to overcrystallization (for rigid planar structures) or intimate mixing (when there are fewer chemical differences between donor and acceptor). Li et al. confirmed that additives can be used to address both L–L and S–L demixing processes in donor:NFA systems using PTB7:PDI and p-DTS(FBTTh2)2:PDI, respectively.[201] The greater and tunable solubility of NFAs will likely widen the selection of deposition solvents[202] and solvent additives.[203] Given similarities in the BHJ components’ chem-ical structures, solvents additives may need to exploit smaller differences in solubility. However, additives in donor:NFA blends can be also be used to manipulate acceptor crystalliza-tion. For example, DIO can prevent overcrystallization of the acceptor PDI by promoting competing crystal growth of the donor p-DTS(FBTTh2)2,[204] while DIO reduces the perfor-mance of PTB7:ITIC by coarsening ITIC domains.[205] The molecular structures of successful NFAs continue to diversify and grow—as the field matures, in situ morphological studies of polymer:NFA and small molecule:NFA blends will bear out or clarify our current mechanistic understanding of solvent additives.

Similarly, an increasing number of ternary blends that incorporate an additional donor and acceptor components are being reported.[206,207] A ternary blend adds an additional dimension to the phase diagram such that the film formation of a ternary can exhibit emergent behavior not observed in the binary blends.[208] Common motivations for adding additional BHJ components are improving light harvesting by extending film absorption, reducing charge recombination through ener-getic offsets, but also includes morphological control.[206,207] Indeed, the optimal BHJ morphology for a given ternary blend depends on the intended function of the additional component, e.g., whether it forms an alloyed phase, part of an interfacial energetic cascade, or a parallel transport pathway.[209,210] The ternary component can also serve as a morphological control agent, reducing the need for solvent additives or synergistically enhancing their effects. For example, Zhang et al. used 10 wt% P3BT to nucleate donor crystallites in p-DTS(FBTTh2)2:PC71BM, improving the PCE from 3.4% to 5.0% without the need for high temperature annealing or solvent additives.[211] The P3BT polymer heterogeneously nucleates donor crystallites into interconnected nanofibers, to similar effect as the insulating solid additive PS;[92,212] however, in this ternary blend, P3BT also contributes to photogeneration and charge transport, increasing the hole mobility by an order of magnitude.[211] Lee et al. accomplished similar results by combining 2 wt% PCDTBT with 0.4% v/v DIO in the same D:A blend, synergisti-cally improving performance from 2.4% (without either addi-tive) to 7.9%.[213] Significantly, the ternary blend morphology was less sensitive to thermal treatment and the concentration of solvent additive, as observed for other binary additive com-binations.[120,214] In films cast from 3% v/v DIO, overcrystalliza-tion could be suppressed by increasing the polymer concentra-tion to 20 wt%—maintaining a PCE of 6.6%, compared to 2.5% without the additional polymer donor. Additionally, Lee et al.

showed that PCDTBT’s ability to nucleate p-DTS(FBTTh2)2 stems from favorable interactions between benzothiadiazole subunits, suggesting future criteria for controlling morphology via ternary components.[213] As a final example, Zhang et al. combined 3% v/v DIO with 15 donor wt% p-DTS(FBTTh2)2 as a sensitizer for PTB7-Th:PC71BM blends to achieve 10.5% PCE.[215] The two donors form an alloyed phase with the more crystalline small molecule templating face-on PTB7-Th crystal-lites that are favorable for vertical charge transport.[215] Both the small molecule and DIO additives promote phase separation via S–L demixing, suppressing L–L demixing processes that lead to reduced efficiency in as-cast PTB7-Th:PC71BM films.[216] As our understanding of these complex blends improves, sol-vent additives will likely be chosen to selectively tune the spatial and temporal development of specific phases.

6. Conclusion

Higher OPV efficiencies have been afforded by design princi-ples derived from fundamental insight into the relationships between molecular structure, processing conditions, BHJ morphology, and device function. Solvent additives have pro-vided tools for manipulating the kinetics of self-assembly and phase separation of the active layer during deposition. In situ X-ray scattering, absorbance, and reflectometry studies of film evolution during solution casting have provided a rich under-standing of the underlying mechanisms that operate during formation of a BHJ. Multiple solvent additives are now being rationally combined or paired with other morphological con-trol strategies to fine tune of BHJ domains simultaneously via different mechanisms, often enabling synergistic effects. However, there are growing concerns over the role of solvent additives in device instability, environmental toxicity, and rel-evance to large-scale manufacturing. Meanwhile, emerging research trends in solution-processed BHJ OPVs will change the prevailing solvent:donor:acceptor paradigm—whether through the use of nonfullerene acceptors, ternary blends, and environmentally sustainable solvents. BHJ morphologies of these emerging fields are unlikely to be optimized as-cast. Sol-vent additive processing will thus continue to be a relevant and effective strategy for affecting desired properties even as our considerations expand beyond that of raw efficiency. Indeed, a greater variety of solvent additives, including nontoxic ones, will likely be identified as the components of the deposition solution increase in diversity and complexity. Furthermore, developments in solvent additive processing that address these additional concerns in OPVs are likely to percolate into related organic semiconductor fields.

AcknowledgementsThis research was supported by the Center for Advanced Organic Photovoltaics, a Multidisciplinary University Research Initiatives (MURI) program under the Department of the Navy, and Office of Naval Research Award Nos. N00014-14-1-0580 and N00014-16-1-2520. This article is part of the Advanced Materials Hall of Fame article series, which recognizes the excellent contributions of leading researchers to the field of materials science.



Adv. Mater. 2018, 30, 1707114

Conflict of InterestThe authors declare no conflict of interest.

Keywordsbulk heterojunctions, morphology, organic photovoltaics, phase separation, solvent additives

Received: December 5, 2017Revised: February 17, 2018

Published online: June 13, 2018

[1] S. B. Darling, F. You, RSC Adv. 2013, 3, 17633.[2] C. J. M. Emmott, J. A. Röhr, M. Campoy-Quiles, T. Kirchartz,

A. Urbina, N. J. Ekins-Daukes, J. Nelson, Energy Environ. Sci. 2015, 8, 1317.

[3] Z. Liu, P. You, S. Liu, F. Yan, ACS Nano 2015, 9, 12026.[4] S. Berny, N. Blouin, A. Distler, H.-J. Egelhaaf, M. Krompiec,

A. Lohr, O. R. Lozman, G. E. Morse, L. Nanson, A. Pron, T. Sauermann, N. Seidler, S. Tierney, P. Tiwana, M. Wagner, H. Wilson, Adv. Sci. 2016, 3, 1500342.

[5] Y. Diao, L. Shaw, Z. Bao, S. C. B. Mannsfeld, Energy Environ. Sci. 2014, 7, 2145.

[6] C. K. Chiang, S. C. Gau, C. R. Fincher, Y. W. Park, A. G. MacDiarmid, A. J. Heeger, Appl. Phys. Lett. 1978, 33, 18.

[7] A. G. MacDiarmid, A. J. Heeger, Synth. Met. 1980, 1, 101.[8] G. Yu, J. Gao, J. C. Hummelen, F. Wudl, A. J. Heeger, Science 1995,

270, 1789.[9] T. A. Skotheim, J. R. Reynolds, Handbook of Conducting Polymers,

Vol. 1, CRC Press, Boca Raton, FL 2007.[10] S. Karg, W. Riess, V. Dyakonov, M. Schwoerer, Synth. Met. 1993,

54, 427.[11] N. S. Sariciftci, L. Smilowitz, A. J. Heeger, F. Wudl, Science 1992,

258, 1474.[12] B. Kraabel, C. H. Lee, D. McBranch, D. Moses, N. S. Sariciftci,

A. J. Heeger, Chem. Phys. Lett. 1993, 213, 389.[13] C. J. Brabec, G. Zerza, G. Cerullo, S. De Silvestri, S. Luzzati,

J. C. Hummelen, S. Sariciftci, Chem. Phys. Lett. 2001, 340, 232.[14] G. A. Chamberlain, Sol. Cells 1983, 8, 47.[15] C. W. Tang, Appl. Phys. Lett. 1986, 48, 183.[16] S. A. Gevorgyan, F. C. Krebs, Chem. Mater. 2008, 20, 4386.[17] K. Nakano, K. Tajima, Adv. Mater. 2017, 29, 1603269.[18] N. D. Treat, A. Varotto, C. J. Takacs, N. Batara, M. Al-Hashimi,

M. J. Heeney, A. J. Heeger, F. Wudl, C. J. Hawker, M. L. Chabinyc, J. Am. Chem. Soc. 2012, 134, 15869.

[19] N. E. Jackson, B. M. Savoie, T. J. Marks, L. X. Chen, M. A. Ratner, J. Phys. Chem. Lett. 2015, 6, 77.

[20] K. Vandewal, S. Albrecht, E. T. Hoke, K. R. Graham, J. Widmer, J. D. Douglas, M. Schubert, W. R. Mateker, J. T. Bloking, G. F. Burkhard, A. Sellinger, J. M. J. Fréchet, A. Amassian, M. K. Riede, M. D. McGehee, D. Neher, A. Salleo, Nat. Mater. 2014, 13, 63.

[21] S. Sweetnam, R. Prasanna, T. M. Burke, J. A. Bartelt, M. D. McGehee, J. Phys. Chem. C 2016, 120, 6427.

[22] G. F. Burkhard, E. T. Hoke, S. R. Scully, M. D. McGehee, Nano Lett. 2009, 9, 4037.

[23] T. M. Burke, M. D. McGehee, Adv. Mater. 2014, 26, 1923.[24] Y. Huang, E. J. Kramer, A. J. Heeger, G. C. Bazan, Chem. Rev. 2014,

114, 7006.[25] X. Jiao, L. Ye, H. Ade, Adv. Energy Mater. 2017, 7, 1700084.[26] N. K. Elumalai, A. Uddin, Energy Environ. Sci. 2016, 9, 391.

[27] B. Ray, M. S. Lundstrom, M. A. Alam, Appl. Phys. Lett. 2012, 100, 013307.

[28] D. Bartesaghi, I. del C. Pérez, J. Kniepert, S. Roland, M. Turbiez, D. Neher, L. J. A. Koster, Nat. Commun. 2015, 6, 7083.

[29] M.-H. Jao, H.-C. Liao, W.-F. Su, J. Mater. Chem. A 2016, 4, 5784.[30] G. F. Burkhard, E. T. Hoke, M. D. McGehee, Adv. Mater. 2010, 22,

3293.[31] F. Padinger, R. s. Rittberger, N. s. Sariciftci, Adv. Funct. Mater.

2003, 13, 85.[32] E. Verploegen, R. Mondal, C. J. Bettinger, S. Sok, M. F. Toney,

Z. Bao, Adv. Funct. Mater. 2010, 20, 3519.[33] E. Verploegen, C. E. Miller, K. Schmidt, Z. Bao, M. F. Toney, Chem.

Mater. 2012, 24, 3923.[34] J. Min, N. S. Güldal, J. Guo, C. Fang, X. Jiao, H. Hu, T. Heumüller,

H. Ade, C. J. Brabec, J. Mater. Chem. A 2017, 5, 18101.[35] G. Li, V. Shrotriya, J. Huang, Y. Yao, T. Moriarty, K. Emery, Y. Yang,

Nat. Mater. 2005, 4, 864.[36] G. Li, Y. Yao, H. Yang, V. Shrotriya, G. Yang, Y. Yang, Adv. Funct.

Mater. 2007, 17, 1636.[37] D. Bharti, V. Raghuwanshi, I. Varun, A. K. Mahato, S. P. Tiwari,

ACS Appl. Mater. Interfaces 2017, 9, 26226.[38] J.-H. Kim, J. H. Park, J. H. Lee, J. S. Kim, M. Sim, C. Shim, K. Cho,

J. Mater. Chem. 2010, 20, 7398.[39] K. Sun, Z. Xiao, E. Hanssen, M. F. G. Klein, H. H. Dam, M. Pfaff,

D. Gerthsen, W. W. H. Wong, D. J. Jones, J. Mater. Chem. A 2014, 2, 9048.

[40] J. Peet, C. Soci, R. C. Coffin, T. Q. Nguyen, A. Mikhailovsky, D. Moses, G. C. Bazan, Appl. Phys. Lett. 2006, 89, 252105.

[41] J. Peet, J. Y. Kim, N. E. Coates, W. L. Ma, D. Moses, A. J. Heeger, G. C. Bazan, Nat. Mater. 2007, 6, 497.

[42] J. T. Rogers, K. Schmidt, M. F. Toney, E. J. Kramer, G. C. Bazan, Adv. Mater. 2011, 23, 2284.

[43] J. K. Lee, W. L. Ma, C. J. Brabec, J. Yuen, J. S. Moon, J. Y. Kim, K. Lee, G. C. Bazan, A. J. Heeger, J. Am. Chem. Soc. 2008, 130, 3619.

[44] B. Russ, A. Glaudell, J. J. Urban, M. L. Chabinyc, R. A. Segalman, Nat. Rev. Mater. 2016, 1, 201650.

[45] W. Yin, M. Dadmun, ACS Nano 2011, 5, 4756.[46] W. Chen, T. Xu, F. He, W. Wang, C. Wang, J. Strzalka, Y. Liu, J. Wen,

D. J. Miller, J. Chen, K. Hong, L. Yu, S. B. Darling, Nano Lett. 2011, 11, 3707.

[47] J. Peet, N. S. Cho, S. K. Lee, G. C. Bazan, Macromolecules 2008, 41, 8655.

[48] C. V. Hoven, X.-D. Dang, R. C. Coffin, J. Peet, T.-Q. Nguyen, G. C. Bazan, Adv. Mater. 2010, 22, E63.

[49] K. Schmidt, C. J. Tassone, J. R. Niskala, A. T. Yiu, O. P. Lee, T. M. Weiss, C. Wang, J. M. J. Fréchet, P. M. Beaujuge, M. F. Toney, Adv. Mater. 2014, 26, 300.

[50] T. L. Nguyen, H. Choi, S.-J. Ko, M. A. Uddin, B. Walker, S. Yum, J.-E. Jeong, M. H. Yun, T. J. Shin, S. Hwang, J. Y. Kim, H. Y. Woo, Energy Environ. Sci. 2014, 7, 3040.

[51] M. A. Uddin, T. H. Lee, S. Xu, S. Y. Park, T. Kim, S. Song, T. L. Nguyen, S. Ko, S. Hwang, J. Y. Kim, H. Y. Woo, Chem. Mater. 2015, 27, 5997.

[52] H. Choi, S.-J. Ko, T. Kim, P.-O. Morin, B. Walker, B. H. Lee, M. Leclerc, J. Y. Kim, A. J. Heeger, Adv. Mater. 2015, 27, 3318.

[53] T. H. Lee, S. Y. Park, B. Walker, S.-J. Ko, J. Heo, H. Y. Woo, H. Choi, J. Y. Kim, RSC Adv. 2017, 7, 7476.

[54] J. Y. Na, M. Kim, Y. D. Park, J. Phys. Chem. C 2017, 121, 13930.[55] J. Roncali, P. Leriche, P. Blanchard, Adv. Mater. 2014, 26, 3821.[56] S. D. Collins, N. A. Ran, M. C. Heiber, T.-Q. Nguyen, Adv. Energy

Mater. 2017, 7, 1602242.[57] Y. Sun, G. C. Welch, W. L. Leong, C. J. Takacs, G. C. Bazan,

A. J. Heeger, Nat. Mater. 2012, 11, 44.[58] T. S. van der Poll, J. A. Love, T.-Q. Nguyen, G. C. Bazan, Adv.

Mater. 2012, 24, 3646.



Adv. Mater. 2018, 30, 1707114

[59] J. K. Park, C. Kim, B. Walker, T.-Q. Nguyen, J. H. Seo, RSC Adv. 2012, 2, 2232.

[60] Q. V. Hoang, S. Rasool, S. Oh, D. V. Vu, D. H. Kim, H. K. Lee, C. E. Song, S. K. Lee, J.-C. Lee, S.-J. Moon, W. S. Shin, J. Mater. Chem. C 2017, 5, 7837.

[61] J. A. Love, S. D. Collins, I. Nagao, S. Mukherjee, H. Ade, G. C. Bazan, T.-Q. Nguyen, Adv. Mater. 2014, 26, 7308.

[62] J. A. Love, C. M. Proctor, J. Liu, C. J. Takacs, A. Sharenko, T. S. van der Poll, A. J. Heeger, G. C. Bazan, T.-Q. Nguyen, Adv. Funct. Mater. 2013, 23, 5019.

[63] T. Salim, L. H. Wong, B. Bräuer, R. Kukreja, Y. L. Foo, Z. Bao, Y. M. Lam, J. Mater. Chem. 2010, 21, 242.

[64] M. Li, W. Zhang, X. Tang, J. Jin, H. Wang, L. Chen, W. Lv, R. Chen, W. Huang, ACS Sustainable Chem. Eng. 2017, 5, 11668.

[65] S. W. Heo, K. W. Song, M. H. Choi, H. S. Oh, D. K. Moon, Sol. Energy Mater. Sol. Cells 2013, 114, 82.

[66] X. Sun, C. Li, L. Huang, R. Xu, J. Ni, H. Cai, J. Li, J. Zhang, Polym. Bull. 2017, 74, 4515.

[67] K. Yao, Y.-X. Xu, X. Wang, F. Li, J. Yuan, RSC Adv. 2015, 5, 93689.

[68] H. I. Kim, M. Kim, C. W. Park, H. U. Kim, H.-K. Lee, T. Park, Chem. Mater. 2017, 29, 6793.

[69] A. Zhang, Q. Wang, R. A. A. Bovee, C. Li, J. Zhang, Y. Zhou, Z. Wei, Y. Li, R. A. J. Janssen, Z. Wang, W. Li, J. Mater. Chem. A 2016, 4, 7736.

[70] T.-Y. Chu, S. Alem, S.-W. Tsang, S.-C. Tse, S. Wakim, J. Lu, G. Dennler, D. Waller, R. Gaudiana, Y. Tao, Appl. Phys. Lett. 2011, 98, 253301.

[71] R. C. Mahadevapuram, J. A. Carr, Y. Chen, S. Bose, K. S. Nalwa, J. W. Petrich, S. Chaudhary, Synth. Met. 2013, 185–186, 115.

[72] K. R. Graham, J. Mei, R. Stalder, J. W. Shim, H. Cheun, F. Steffy, F. So, B. Kippelen, J. R. Reynolds, ACS Appl. Mater. Interfaces 2011, 3, 1210.

[73] J. Mei, K. R. Graham, R. Stalder, J. R. Reynolds, Org. Lett. 2010, 12, 660.

[74] J. A. Carr, K. S. Nalwa, R. Mahadevapuram, Y. Chen, J. Anderegg, S. Chaudhary, ACS Appl. Mater. Interfaces 2012, 4, 2831.

[75] F. Machui, C. J. Brabec, in Semiconducting Polymer Composites (Ed: X. Yang), Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim, Germany 2012, pp. 1–38.

[76] A. F. M. Barton, Chem. Rev. 1975, 75, 731.[77] C. Della Volpe, D. Maniglio, M. Brugnara, S. Siboni, M. Morra, J.

Colloid Interface Sci. 2004, 271, 434.[78] S. Siboni, C. Della Volpe, D. Maniglio, M. Brugnara, J. Colloid Inter-

face Sci. 2004, 271, 454.[79] C. M. Hansen, Hansen Solubility Parameters: A User’s Handbook,

2nd ed., CRC Press, Boca Raton, FL 2007.[80] C. M. Hansen, A. L. Smith, Carbon 2004, 42, 1591.[81] B. Walker, A. Tamayo, D. T. Duong, X.-D. Dang, C. Kim,

J. Granstrom, T.-Q. Nguyen, Adv. Energy Mater. 2011, 1, 221.[82] K. R. Graham, P. M. Wieruszewski, R. Stalder, M. J. Hartel, J. Mei,

F. So, J. R. Reynolds, Adv. Funct. Mater. 2012, 22, 4801.[83] Y. Choi, G. Kim, H. Kim, S. H. Lee, S. Kwon, J. Kim, K. Lee, Nano

Energy 2016, 30, 200.[84] S. Kwon, H. Kang, J.-H. Lee, J. Lee, S. Hong, H. Kim, K. Lee, Adv.

Energy Mater. 2017, 7, 1601496.[85] E. F. Manley, J. Strzalka, T. J. Fauvell, N. E. Jackson, M. J. Leonardi,

N. D. Eastham, T. J. Marks, L. X. Chen, Adv. Mater. 2017, 1703933.[86] F. Machui, P. Maisch, I. Burgués-Ceballos, S. Langner, J. Krantz,

T. Ameri, C. J. Brabec, ChemPhysChem 2015, 16, 1275.[87] J. Wang, Z. Liang, ACS Appl. Mater. Interfaces 2016, 8, 22418.[88] A. T. Kleinschmidt, S. E. Root, D. J. Lipomi, J. Mater. Chem. A

2016, 5, 11396.[89] S. Zhang, L. Ye, J. Hou, Adv. Energy Mater. 2016, 6, 1502529.

[90] L. A. Perez, K. W. Chou, J. A. Love, T. S. van der Poll, D.-M. Smilgies, T.-Q. Nguyen, E. J. Kramer, A. Amassian, G. C. Bazan, Adv. Mater. 2013, 25, 6380.

[91] L. A. Perez, J. T. Rogers, M. A. Brady, Y. Sun, G. C. Welch, K. Schmidt, M. F. Toney, H. Jinnai, A. J. Heeger, M. L. Chabinyc, G. C. Bazan, E. J. Kramer, Chem. Mater. 2014, 26, 6531.

[92] C. McDowell, M. Abdelsamie, K. Zhao, D.-M. Smilgies, G. C. Bazan, A. Amassian, Adv. Energy Mater. 2015, 5, 1501121.

[93] M. Abdelsamie, N. D. Treat, K. Zhao, C. McDowell, M. A. Burgers, R. Li, D.-M. Smilgies, N. Stingelin, G. C. Bazan, A. Amassian, Adv. Mater. 2015, 27, 7285.

[94] M. Kim, J. Lee, S. B. Jo, D. H. Sin, H. Ko, H. Lee, S. G. Lee, K. Cho, J. Mater. Chem. A 2016, 4, 15522.

[95] S. Kouijzer, J. J. Michels, M. van den Berg, V. S. Gevaerts, M. Turbiez, M. M. Wienk, R. A. J. Janssen, J. Am. Chem. Soc. 2013, 135, 12057.

[96] J. D. Perea, S. Langner, M. Salvador, B. Sanchez-Lengeling, N. Li, C. Zhang, G. Jarvas, J. Kontos, A. Dallos, A. Aspuru-Guzik, C. J. Brabec, J. Phys. Chem. C 2017, 121, 18153.

[97] J.-H. Kim, A. Gadisa, C. Schaefer, H. Yao, B. R. Gautam, N. Balar, M. Ghasemi, I. Constantinou, F. So, B. T. O’Connor, K. Gundogdu, J. Hou, H. Ade, J. Mater. Chem. A 2017, 5, 13176.

[98] S. Nilsson, A. Bernasik, A. Budkowski, E. Moons, Macromolecules 2007, 40, 8291.

[99] L. J. Richter, D. M. DeLongchamp, A. Amassian, Chem. Rev. 2017, 117, 6332.

[100] A. Sharenko, N. D. Treat, J. A. Love, M. F. Toney, N. Stingelin, T.-Q. Nguyen, J. Mater. Chem. A 2014, 2, 15717.

[101] M. J. A. Nekuda, N. D. Treat, M. Abdelsamie, L. Yu, R. Li, D.-M. Smilgies, A. Amassian, C. J. Hawker, M. L. Chabinyc, N. Stingelin, Org. Photonics Photovoltaics 2014, 2, 59.

[102] P. Westacott, N. D. Treat, J. Martin, J. H. Bannock, J. C. de Mello, M. Chabinyc, A. B. Sieval, J. J. Michels, N. Stingelin, J. Mater. Chem. A 2017, 5, 2689.

[103] J. Yin, W. Zhou, L. Zhang, Y. Xie, Z. Yu, J. Shao, W. Ma, J. Zeng, Y. Chen, Macromol. Rapid Commun. 2017, 38, 1700428.

[104] L. Li, W. Niu, X. Zhao, X. Yang, S. Chen, Sci. Adv. Mater. 2015, 7, 2021.[105] F. Buss, B. Schmidt-Hansberg, M. Sanyal, C. Munuera, P. Scharfer,

W. Schabel, E. Barrena, Macromolecules 2016, 49, 4867.[106] N. S. Güldal, T. Kassar, M. Berlinghof, T. Unruh, C. J. Brabec, J.

Mater. Res. 2017, 32, 1855.[107] F. C. Krebs, Sol. Energy Mater. Sol. Cells 2009, 93, 394.[108] J. J. van Franeker, M. Turbiez, W. Li, M. M. Wienk, R. A. J. Janssen,

Nat. Commun. 2015, 6, 6229.[109] J. A. Bartelt, J. D. Douglas, W. R. Mateker, A. El Labban,

C. J. Tassone, M. F. Toney, J. M. J. Fréchet, P. M. Beaujuge, M. D. McGehee, Adv. Energy Mater. 2014, 4, 1.

[110] W. Kim, J. K. Kim, E. Kim, T. K. Ahn, D. H. Wang, J. H. Park, J. Phys. Chem. C 2015, 119, 5954.

[111] S. Guo, E. M. Herzig, A. Naumann, G. Tainter, J. Perlich, P. Müller-Buschbaum, J. Phys. Chem. B 2014, 118, 344.

[112] J. J. van Franeker, G. H. L. Heintges, C. Schaefer, G. Portale, W. Li, M. M. Wienk, P. van der Schoot, R. A. J. Janssen, J. Am. Chem. Soc. 2015, 137, 11783.

[113] H.-C. Liao, C.-C. Ho, C.-Y. Chang, M.-H. Jao, S. B. Darling, W.-F. Su, Mater. Today 2013, 16, 326.

[114] F. Liu, Y. Gu, C. Wang, W. Zhao, D. Chen, A. L. Briseno, T. P. Russell, Adv. Mater. 2012, 24, 3947.

[115] J. Rivnay, S. C. B. Mannsfeld, C. E. Miller, A. Salleo, M. F. Toney, Chem. Rev. 2012, 112, 5488.

[116] D. Chen, A. Nakahara, D. Wei, D. Nordlund, T. P. Russell, Nano Lett. 2011, 11, 561.

[117] L. J. Richter, D. M. DeLongchamp, F. A. Bokel, S. Engmann, K. W. Chou, A. Amassian, E. Schaible, A. Hexemer, Adv. Energy Mater. 2015, 5, 1400975.



[118] J. T. Rogers, K. Schmidt, M. F. Toney, G. C. Bazan, E. J. Kramer, J. Am. Chem. Soc. 2012, 134, 2884.

[119] Y. Gu, C. Wang, T. P. Russell, Adv. Energy Mater. 2012, 2, 683.[120] A. Sharenko, M. Kuik, M. F. Toney, T.-Q. Nguyen, Adv. Funct.

Mater. 2014, 24, 3543.[121] P. Mokarian-Tabari, M. Geoghegan, J. R. Howse, S. Y. Heriot,

R. L. Thompson, R. a. L. Jones, Eur. Phys. J. E 2010, 33, 283.[122] Q. Sun, F. Zhang, Q. An, M. Zhang, X. Ma, J. Zhang, ACS Appl.

Mater. Interfaces 2017, 9, 8863.[123] X. Liao, R. Lv, L. Chen, Y. Chen, Phys. Chem. Chem. Phys. 2017, 19,

10581.[124] J. S. Moon, C. J. Takacs, Y. Sun, A. J. Heeger, Nano Lett. 2011, 11,

1036.[125] L. Ye, S. Zhang, W. Ma, B. Fan, X. Guo, Y. Huang, H. Ade, J. Hou,

Adv. Mater. 2012, 24, 6335.[126] D. H. Wang, A. K. K. Kyaw, J.-R. Pouliot, M. Leclerc, A. J. Heeger,

Adv. Energy Mater. 2014, 4, 1300835.[127] Q. Wan, X. Guo, Z. Wang, W. Li, B. Guo, W. Ma, M. Zhang, Y. Li,

Adv. Funct. Mater. 2016, 26, 6635.[128] Y.-S. Jung, J.-S. Yeo, N.-K. Kim, S. Lee, D.-Y. Kim, ACS Appl. Mater.

Interfaces 2016, 8, 30372.[129] C. Wang, L. Jiang, W. Hu, in Organic Optoelectronics (Ed: W. Hu),

Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim, Germany 2013, pp. 95–170.

[130] J. Smith, R. Hamilton, I. McCulloch, N. Stingelin-Stutzmann, M. Heeney, D. D. C. Bradley, T. D. Anthopoulos, J. Mater. Chem. 2010, 20, 2562.

[131] J. W. Ward, Z. A. Lamport, O. D. Jurchescu, ChemPhysChem 2015, 16, 1118.

[132] H. Sirringhaus, Adv. Mater. 2014, 26, 1319.[133] J.-F. Chang, B. Sun, D. W. Breiby, M. M. Nielsen, T. I. Sölling,

M. Giles, I. McCulloch, H. Sirringhaus, Chem. Mater. 2004, 16, 4772.

[134] D. M. DeLongchamp, B. M. Vogel, Y. Jung, M. C. Gurau, C. A. Richter, O. A. Kirillov, J. Obrzut, D. A. Fischer, S. Sambasivan, L. J. Richter, E. K. Lin, Chem. Mater. 2005, 17, 5610.

[135] H. Yang, T. J. Shin, L. Yang, K. Cho, C. Y. Ryu, Z. Bao, Adv. Funct. Mater. 2005, 15, 671.

[136] H. Yang, S. W. LeFevre, C. Y. Ryu, Z. Bao, Appl. Phys. Lett. 2007, 90, 172116.

[137] M. Brinkmann, L. Hartmann, L. Biniek, K. Tremel, N. Kayunkid, Macromol. Rapid Commun. 2014, 35, 9.

[138] Y. Yuan, G. Giri, A. L. Ayzner, A. P. Zoombelt, S. C. B. Mannsfeld, J. Chen, D. Nordlund, M. F. Toney, J. Huang, Z. Bao, Nat. Commun. 2014, 5, 4005.

[139] H.-R. Tseng, L. Ying, B. B. Y. Hsu, L. A. Perez, C. J. Takacs, G. C. Bazan, A. J. Heeger, Nano Lett. 2012, 12, 6353.

[140] H.-R. Tseng, H. Phan, C. Luo, M. Wang, L. A. Perez, S. N. Patel, L. Ying, E. J. Kramer, T.-Q. Nguyen, G. C. Bazan, A. J. Heeger, Adv. Mater. 2014, 26, 2993.

[141] M. Pandey, S. S. Pandey, S. Nagamatsu, S. Hayase, W. Takashima, Org. Electron. 2017, 43, 240.

[142] M. Ichikawa, K. Yamamura, H.-G. Jeon, M. Nakajima, Y. Taniguchi, J. Appl. Phys. 2011, 109, 054504.

[143] D. Bharti, S. P. Tiwari, Synth. Met. 2016, 215, 1.[144] F. Liu, C. Wang, J. K. Baral, L. Zhang, J. J. Watkins, A. L. Briseno,

T. P. Russell, J. Am. Chem. Soc. 2013, 135, 19248.[145] T. K. An, I. Kang, H. Yun, H. Cha, J. Hwang, S. Park, J. Kim,

Y. J. Kim, D. S. Chung, S.-K. Kwon, Y.-H. Kim, C. E. Park, Adv. Mater. 2013, 25, 7003.

[146] Y. J. Park, J. H. Seo, W. Elsawy, B. Walker, S. Cho, J.-S. Lee, J. Mater. Chem. C 2015, 3, 5951.

[147] J. K. Park, B. Walker, J. H. Seo, ACS Appl. Mater. Interfaces 2013, 5, 4575.

[148] M. Nikolka, I. Nasrallah, B. Rose, M. K. Ravva, K. Broch, A. Sadhanala, D. Harkin, J. Charmet, M. Hurhangee, A. Brown, S. Illig, P. Too, J. Jongman, I. McCulloch, J.-L. Bredas, H. Sirringhaus, Nat. Mater. 2017, 16, 356.

[149] L. Ye, Y. Jing, X. Guo, H. Sun, S. Zhang, M. Zhang, L. Huo, J. Hou, J. Phys. Chem. C 2013, 117, 14920.

[150] Y. Chen, Y. Zhao, Z. Liang, Energy Environ. Sci. 2015, 8, 401.[151] G. Chen, W. Xu, D. Zhu, J. Mater. Chem. C 2017, 5, 4350.[152] R. Kroon, D. A. Mengistie, D. Kiefer, J. Hynynen, J. D. Ryan, L. Yu,

C. Müller, Chem. Soc. Rev. 2016, 45, 6147.[153] Z. Zhu, C. Liu, F. Jiang, J. Xu, E. Liu, Synth. Met. 2017, 225, 31.[154] J. Y. Kim, J. H. Jung, D. E. Lee, J. Joo, Synth. Met. 2002, 126, 311.[155] G.-H. Kim, L. Shao, K. Zhang, K. P. Pipe, Nat. Mater. 2013, 12, 719.[156] Q. Wei, M. Mukaida, Y. Naitoh, T. Ishida, Adv. Mater. 2013, 25, 2831.[157] I. Petsagkourakis, E. Pavlopoulou, G. Portale, B. A. Kuropatwa,

S. Dilhaire, G. Fleury, G. Hadziioannou, Sci. Rep. 2016, 6, 30501.[158] F. Jiang, L. Wang, C. Li, X. Wang, Y. Hu, H. Liu, H. Yang, F. Zhao,

J. Xu, J. Polym. Res. 2017, 24, 68.[159] Z. Zhu, C. Liu, H. Shi, Q. Jiang, J. Xu, F. Jiang, J. Xiong, E. Liu, J.

Polym. Sci., Part B: Polym. Phys. 2015, 53, 885.[160] Y. Chen, M. He, J. Peng, Y. Sun, Z. Liang, Adv. Sci. 2016, 3,

1500392.[161] A. Sharenko, M. F. Toney, J. Am. Chem. Soc. 2016, 138, 463.[162] T. Li, Y. Pan, Z. Wang, Y. Xia, Y. Chen, W. Huang, J. Mater. Chem. A

2017, 5, 12602.[163] T. Jeon, S. J. Kim, J. Yoon, J. Byun, H. R. Hong, T.-W. Lee, J.-S. Kim,

B. Shin, S. O. Kim, Adv. Energy Mater. 2017, 7, 1602596.[164] P.-W. Liang, C.-Y. Liao, C.-C. Chueh, F. Zuo, S. T. Williams,

X.-K. Xin, J. Lin, A. K.-Y. Jen, Adv. Mater. 2014, 26, 3748.[165] C.-C. Chueh, C.-Y. Liao, F. Zuo, S. T. Williams, P.-W. Liang,

A. K.-Y. Jen, J. Mater. Chem. A 2015, 3, 9058.[166] P. Mao, Q. Zhou, Z. Jin, H. Li, J. Wang, ACS Appl. Mater. Interfaces

2016, 8, 23837.[167] Q. Wei, D. Yang, Z. Yang, X. Ren, Y. Liu, J. Feng, X. Zhu, S. (Frank)

Liu, RSC Adv. 2016, 6, 56807.[168] X. Song, W. Wang, P. Sun, W. Ma, Z.-K. Chen, Appl. Phys. Lett.

2015, 106, 033901.[169] L. Li, Y. Chen, Z. Liu, Q. Chen, X. Wang, H. Zhou, Adv. Mater.

2016, 28, 9862.[170] J. Mo, C. Zhang, J. Chang, H. Yang, H. Xi, D. Chen, Z. Lin, G. Lu,

J. Zhang, Y. Hao, J. Mater. Chem. A 2017, 5, 13032.[171] J. Huang, S. Tan, P. Lund, H. Zhou, Energy Environ. Sci. 2017, 10,

2284.[172] X. Gong, M. Li, X.-B. Shi, H. Ma, Z.-K. Wang, L.-S. Liao, Adv. Funct.

Mater. 2015, 25, 6671.[173] W. Zhao, D. Qian, S. Zhang, S. Li, O. Inganäs, F. Gao, J. Hou, Adv.

Mater. 2016, 28, 4734.[174] J. Zhao, Y. Li, G. Yang, K. Jiang, H. Lin, H. Ade, W. Ma, H. Yan,

Nat. Energy 2016, 1, 15027.[175] J.-D. Chen, C. Cui, Y.-Q. Li, L. Zhou, Q.-D. Ou, C. Li, Y. Li,

J.-X. Tang, Adv. Mater. 2015, 27, 1035.[176] R. Po, A. Bernardi, A. Calabrese, C. Carbonera, G. Corso,

A. Pellegrino, Energy Environ. Sci. 2014, 7, 925.[177] H. Kang, G. Kim, J. Kim, S. Kwon, H. Kim, K. Lee, Adv. Mater.

2016, 28, 7821.[178] L. Chang, H. W. A. Lademann, J.-B. Bonekamp, K. Meerholz,

A. J. Moulé, Adv. Funct. Mater. 2011, 21, 1779.[179] A. Tournebize, A. Rivaton, H. Peisert, T. Chassé, J. Phys. Chem. C

2015, 119, 9142.[180] B. J. Tremolet de Villers, K. A. O’Hara, D. P. Ostrowski,

P. H. Biddle, S. E. Shaheen, M. L. Chabinyc, D. C. Olson, N. Kopidakis, Chem. Mater. 2016, 28, 876.

[181] X. Zhao, J. Zhao, R. Wu, D. Liu, G. Wang, P. Li, L. Chen, L. Zhu, B. Ding, Q. Song, Org. Electron. 2017, 49, 226.

Adv. Mater. 2018, 30, 1707114



Adv. Mater. 2018, 30, 1707114

[182] W.-H. Chang, L. Meng, L. Dou, J. You, C.-C. Chen, Y. (Michael) Yang, E. P. Young, G. Li, Y. Yang, Macromolecules 2015, 48, 562.

[183] H. Zhang, H. Yao, W. Zhao, L. Ye, J. Hou, Adv. Energy Mater. 2016, 6, 1502177.

[184] J. W. Jung, J. W. Jo, C.-C. Chueh, F. Liu, W. H. Jo, T. P. Russell, A. K.-Y. Jen, Adv. Mater. 2015, 27, 3310.

[185] K. W. Fried, K. K. Rozman, K.-H. Summer, S. Halbach, H. Kappus, H. Greim, P. J. A. Borm, G. H. Degen, J. W. Owens, W. Dekant, M. W. Anders, L. Szinicz, T. Zilker, in Toxicology and Risk Assessment: A Comprehensive Introduction (Eds: H. Greim, R. Snyder), John Wiley & Sons, Ltd., Chichester, UK 2008, pp. 513–655.

[186] N. Espinosa, Y.-S. Zimmermann, G. A. dos R. Benatto, M. Lenz, F. C. Krebs, Energy Environ. Sci. 2016, 9, 1674.

[187] Y.-S. Zimmermann, A. Schäffer, C. Hugi, K. Fent, P. F.-X. Corvini, M. Lenz, Environ. Int. 2012, 49, 128.

[188] J. Kalowekamo, E. Baker, Sol. Energy 2009, 83, 1224.[189] J. Jean, P. R. Brown, R. L. Jaffe, T. Buonassisi, V. Bulovic, Energy

Environ. Sci. 2015, 8, 1200.[190] C. Capello, U. Fischer, K. Hungerbühler, Green Chem. 2007, 9, 927.[191] C. McDowell, G. C. Bazan, Curr. Opin. Green Sustainable Chem.

2017, 5, 49.[192] X. Chen, X. Liu, M. A. Burgers, Y. Huang, G. C. Bazan, Angew.

Chem., Int. Ed. 2014, 53, 14378.[193] Y. Chen, Y. Cui, S. Zhang, J. Hou, Polym. Chem. 2015, 6, 4089.[194] W. Zhao, L. Ye, S. Zhang, M. Sun, J. Hou, J. Mater. Chem. A 2015,

3, 12723.[195] S. Lee, J. Kong, K. Lee, Adv. Energy Mater. 2016, 6, 1600970.[196] J. Griffin, A. J. Pearson, N. W. Scarratt, T. Wang, A. D. F. Dunbar,

H. Yi, A. Iraqi, A. R. Buckley, D. G. Lidzey, Org. Electron. 2015, 21, 216.

[197] C. B. Nielsen, S. Holliday, H.-Y. Chen, S. J. Cryer, I. McCulloch, Acc. Chem. Res. 2015, 48, 2803.

[198] S. Li, W. Liu, C.-Z. Li, M. Shi, H. Chen, Small 2017, 13, 1701120.[199] S. Li, Z. Zhang, M. Shi, C.-Z. Li, H. Chen, Phys. Chem. Chem. Phys.

2017, 19, 3440.

[200] S. M. McAfee, J. M. Topple, I. G. Hill, G. C. Welch, J. Mater. Chem. A 2015, 3, 16393.

[201] M. Li, J. Liu, X. Cao, K. Zhou, Q. Zhao, X. Yu, R. Xing, Y. Han, Phys. Chem. Chem. Phys. 2014, 16, 26917.

[202] G. E. Park, S. Choi, S. Y. Park, D. H. Lee, M. J. Cho, D. H. Choi, Adv. Energy Mater. 2017, 7, 1700566.

[203] W. Zhao, S. Zhang, J. Hou, Sci. China Chem. 2016, 59, 1574.[204] A. Sharenko, D. Gehrig, F. Laquai, T.-Q. Nguyen, Chem. Mater.

2014, 26, 4109.[205] X. Song, N. Gasparini, D. Baran, Adv. Electron. Mater. 2017, 2,

1700358.[206] B. M. Savoie, S. Dunaisky, T. J. Marks, M. A. Ratner, Adv. Energy

Mater. 2015, 5, 1400891.[207] L. Lu, M. A. Kelly, W. You, L. Yu, Nat. Photonics 2015, 9, 491.[208] N. Li, F. Machui, D. Waller, M. Koppe, C. J. Brabec, Sol. Energy

Mater. Sol. Cells 2011, 95, 3465.[209] H.-C. Liao, P.-H. Chen, R. P. H. Chang, W.-F. Su, Polymers 2014, 6,

2784.[210] L. Zhang, W. Ma, Chin. J. Polym. Sci. 2017, 35, 184.[211] L. Zhang, M. Pu, W. Zhou, X. Hu, Y. Zhang, Y. Xie, B. Liu, Y. Chen,

J. Phys. Chem. C 2015, 119, 23310.[212] Y. Huang, W. Wen, S. Mukherjee, H. Ade, E. J. Kramer, G. C. Bazan,

Adv. Mater. 2014, 26, 4168.[213] H.-S. Lee, H. Ahn, J. Woong Jo, B. Kim, H. Jung Son, J. Mater.

Chem. A 2016, 4, 18383.[214] Y. J. Kim, S. Ahn, D. H. Wang, C. E. Park, Sci. Rep. 2015, 5, 18024.[215] J. Zhang, Y. Zhang, J. Fang, K. Lu, Z. Wang, W. Ma, Z. Wei, J. Am.

Chem. Soc. 2015, 137, 8176.[216] J. J. van Franeker, D. Westhoff, M. Turbiez, M. M. Wienk,

V. Schmidt, R. A. J. Janssen, Adv. Funct. Mater. 2014, 25, 855.[217] J. A. Love, I. Nagao, Y. Huang, M. Kuik, V. Gupta, C. J. Takacs,

J. E. Coughlin, L. Qi, T. S. van der Poll, E. J. Kramer, A. J. Heeger, T.-Q. Nguyen, G. C. Bazan, J. Am. Chem. Soc. 2014, 136, 3597.

[218] V. Gupta, A. K. K. Kyaw, D. H. Wang, S. Chand, G. C. Bazan, A. J. Heeger, Sci. Rep. 2013, 3, 1965.

本文献由“学霸图书馆-文献云下载”收集自网络，仅供学习交流使用。

学霸图书馆（www.xuebalib.com）是一个“整合众多图书馆数据库资源，

提供一站式文献检索和下载服务”的24 小时在线不限IP

图书馆。

图书馆致力于便利、促进学习与科研，提供最强文献下载服务。

图书馆导航：

图书馆首页文献云下载图书馆入口外文数据库大全疑难文献辅助工具

http://www.xuebalib.com/cloud/

http://www.xuebalib.com/

http://www.xuebalib.com/cloud/


http://www.xuebalib.com/vip.html

http://www.xuebalib.com/db.php

http://www.xuebalib.com/zixun/2014-08-15/44.html


Documents

Solvent Additives: Key Morphology-Directing Agents for ...download.xuebalib.com/4rm4AGSsHkOv.pdf(KAUST), Saudi Arabia, in 2016. He currently works as a postdoctoral scholar in the