Balancing Crystal Size in Small-Molecule Nonfullerene Solar Cells through Fine-Tuning the Film-Forming Kinetics to Fabricate Interpenetrating Network.

The nanoscale interpenetrating network of active layer plays a key role in determining the exciton dissociation and charge transport in all small-molecule nonfullerene solar cells (AS-NFSCs). However, fabricating interpenetrating networks in all small-molecule blends remains a critical hurdle due to the uncontrolled crystallization behavior of small molecules. In this study, we proposed that the balanced crystal size between the donor and the acceptor is an essential prerequisite to construct optimal interpenetrating networks. We also provided a solvent additive strategy to reduce the gap of crystal size between the donor and the acceptor in S-TR:ITIC all small-molecule blend system through manipulating the solution state and film-forming kinetics. As a result, the crystal size of S-TR decreased and the crystal size of ITIC increased, leading to nanoscale interpenetrating networks. This optimized morphology improved the exciton dissociation efficiency and suppressed the bimolecular recombination, achieving almost double power conversion efficiency compared to the reference device. This work demonstrates that manipulation of the balanced crystal size of donor and acceptor may be a key to further boost the efficiency of AS-NFSCs.

Introduction

All small-molecule nonfullerene solar cells (AS-NFSCs), based on the binary blend of small-molecule donors (SMDs) and nonfullerene small-molecule acceptors (NFAs), are of interest due to lots of virtues over conventional polymer/fullerene devices including easily tunable energy levels, well-defined chemical structure, high purity, and stable batch quality.[1−5] Recently, the power conversion efficiency (PCE) of AS-NFSCs have reached ∼10%, with high short-circuit current density and open-circuit voltage.[4] However, as far as we know, the PCE of the state-of-the-art AS-NFSCs is still lower than those of their polymeric counterparts.[6−8] The restriction to obtain high device performance may be the lack of optimal phase separation. As we know, the ideal morphology of bulk heterojunction (BHJ) should involve interpenetrating network of phase-separated domains with large interfacial areas between donor and acceptor, and the dimension should be comparable to the exciton diffusion length.[9−11] However, in comparison with the case of polymer/fullerene or polymer/polymer-based solar cells, achieving a bicontinuous phase structure in AS-NFSCs is very challenging due to narrower processing windows because small molecules are more sensitive to casting conditions.[4,12,13]

Typically, the phase-separation process occurs either by a spinodal decomposition mechanism or nucleation and growth mechanism. The bicontinuous phase separation structures could be selectively controlled by spinodal decomposition in the blend systems with dynamic symmetry.[14−16] However, for the blend systems contained conjugated molecule, the phase separation dynamic process is usually prone to be asymmetric due to the large difference between intermolecular interaction and/or crystallinity of the two components, resulting in a large number of complex and uncontrollable phase morphologies.[16,17] Furthermore, owing to the strong π–π interaction among donor and acceptor molecules, the phase-separation process usually is accompanied by crystallization. When phase separation and crystallization occur simultaneously, the final morphology is mainly determined by the competition of crystallization between the donor and the acceptor.[16,18] Fortunately, regulating the crystallization behavior of both donor and acceptor is an efficient way to control the morphology in crystallization/crystallization blend system.

It is known that the balanced crystallization between donor and acceptor is an important premise to fabricating bicontinuous phase-separation structures or interpenetration networks. According to this principle, the polymer donors with high crystallinity appear to match better with fullerene acceptors due to the easy crystallization of fullerene. However, for the NFAs (usually, NFAs have low crystallinity compared to fullerene), polymer with weak crystallinity should be selected as the donor to achieve good device performance. For example, PffBT4T-2OD (high crystallinity) can yield a PCE of 10.8% when blended with fullerenes. However, blending PffBT4T-2OD with SFPDI2 (NFAs) could only result in a low PCE of ∼3%. Alternatively, great enhancement in PCE can be achieved by reducing the crystallinity of the polymer.[19−21] Similar phenomenon can be found in all the polymer blend systems as well. Ma et al. synthesized a series of donor–acceptor polymers with different crystallinity by introducing different building blocks and side chains. They found that when the crystallinity of the donor and acceptor were resemble, the interpenetrating network could be obtained and a PCE of 6.0% could be achieved.[22]

Although the works about morphology control in AS-NFSCs blend systems are rarely reported, and how the crystallinity of both donor and acceptor affects the phase separation structure of the blend is also obscure.[4,23] Furthermore, the variation trend in the crystallinity of the donor and acceptor is prone to follow the same trend, i.e., their crystallinities are simultaneously enhanced or reduced by using the traditional processing strategies, such as thermal annealing[23] and solvent vapor annealing.[4] There is lack of novel processing strategies to independently control the crystallinity of SMDs and NFAs, for example, suppressing the crystallinity of one component and enhancing the crystallinity of the other component at the same time.

In this work, we proposed that the balanced crystal size between the donor and the acceptor is an important prerequisite to construct the interpenetrating network for AS-NFSCs. Additionally, the independent control of the crystallinity for the donor and the acceptor was also realized by tuning the solution state and film-forming kinetics. Here, BDTT-S-TR (S-TR) was selected as the donor and ITIC was chosen as the acceptor. Due to the large difference in the crystal size between S-TR and ITIC in the blend system, i.e., the crystal size of S-TR (LD) is 16 nm and the crystal size of ITIC (LA) is 0, the phase separation was dominated by the crystallization of S-TR, thus forming a grain-like phase-separation structure. To balance the crystal size, the solvent additive chloronaphthalene (CN) was well selected: on one hand, the crystallinity of S-TR was suppressed due to the increased nucleation barrier in the solution, resulting in a decreased LD (13 nm). On the other hand, the crystallization of ITIC was retarded during the film-forming process, which reduced the disturbance of S-TR on the self-organization of ITIC, leading to an enlarged LA (9.2 nm). The balanced crystal size between S-TR and ITIC induced the formation of interpenetrating network. The optimized morphology facilitates the dissociation of exciton and suppresses the bimolecular recombination during charge transport process, leading to an improved device performance. Our work raised the principle for how to optimize the morphology of AS-NFSCs and also provided a method to individually control the crystallinity of the donor and the acceptor, which could extend the ideal pair of SMDs and NFAs.

Results and Discussion

Due to the tremendous difference in the crystallinity between S-TR and ITIC, the morphology of pristine S-TR:ITIC blend film is dominated by the crystallization of S-TR, which suppresses the aggregation of ITIC, leading to a grain-like phase-separation structure. This structure is not suited for the exciton dissociation in S-TR phase, and also leads to a poor charge transport in the ITIC phase. To fabricate an interpenetrating structure, a balanced crystal size between S-TR and ITIC should be obtained. Here, we adopted CN as a solvent additive to decrease the crystal size of S-TR and enhance the crystal size of ITIC because of the optimized solution state and film-forming kinetics, resulting in an optimal interpenetrating network structure.

Balancing Crystal Size To Fabricate Interpenetrating Network

Herein, we adopted a small-molecule material S-TR with high crystallinity as the donor and ITIC with low crystallinity as the acceptor; the molecular structures and energy levels are displayed in Figure A. The UV–vis absorption spectrum of S-TR:ITIC film is shown in Figure B. The blend film has a very broad absorption from 400 to 780 nm. The absorption around 420–650 nm is mainly attributed to S-TR and the absorption around 650–780 nm is mainly caused by ITIC (Figure S1A).[8,24] After adding CN, two obvious differences appears. First, the absorption shoulder peak centered at 586 nm gradually blueshifts to 575 nm, indicating a decreased aggregation of S-TR. Second, the intensity of absorption shoulder peak around 706 nm gets stronger than the film without CN, which means increased crystallinity of ITIC. The changed aggregated states were also confirmed by photoluminescence (PL) spectroscopy. According to the PL spectroscopy of the neat films (Figure S1B), it is clear that the peaks centered at 701 and 770 nm are the signals of S-TR and ITIC, respectively. As shown in Figure C, the emission intensity of S-TR decreases, whereas that of ITIC increases, suggesting that the opposite aggregated state changes in S-TR and ITIC after adding CN. Usually, the changed trend of aggregated state for donor and acceptor in the NFSM-OSC blend system is always similar when the film undergoes the traditional post-treatment.[4,25] The opposite change tendency we observed in this experiment is rarely reported before, and we will discuss the mechanism in detail in the following sections.

Figure 1

(A) Molecular structures and energy levels of S-TR and ITIC. (B) UV–vis absorption spectra and (C) PL spectroscopy of S-TR:ITIC blend films processed without and with different contents of CN.

The changed aggregated states are usually induced by crystallization, so we characterized the molecular packing and crystal texture by grazing incidence wide-angle X-ray scattering (GIWAXS) as shown in Figure .[26] For the pure S-TR film, not only a strong (100) diffraction signal located at 0.32 Å–1 along the vertical (q) axis, but the (200), (300), and (400) diffraction patterns could also be observed (Figure A), revealing the highly crystalline structure on lamella stacking and a strong preference for edge-on orientation. For the ITIC film, a weak lamellar (100) peak located at 0.50 Å–1 and a π–π stacking (010) diffraction signal located at 1.50 Å–1 were observed along the qz axis as shown in Figure B, which indicates that ITIC has a weak crystallinity and adopts both edge-on and face-on orientation in the film.[27]

Figure 2

GIWAXS patterns of (A) neat S-TR, (B) neat ITIC films, and S-TR:ITIC blend films processed without CN (C), with 1% CN (D), 3% CN (E), and 5% CN (F).

For the S-TR:ITIC blend film, the GIWAXS (Figure C) shows obvious diffraction patterns of S-TR, whereas the diffraction of ITIC could not be observed. It indicates that S-TR crystallized in a similar manner as in the neat S-TR film in the blend system. However, the crystallization of ITIC is totally inhabited after blending with S-TR. The crystal sizes for both S-TR and ITIC were calculated as well (Figure S2). Ade et al. have demonstrated that the crystal size affects the exciton dissociation and charge transport. Hence, the deep analysis of crystal size could provide additional details about the relationship of morphology performance at a nanolength scale.[28,29] As shown in Table , the crystal size of S-TR (LD) is 16.4 nm, whereas the crystal diffraction signal of ITIC was hardly observed and the crystal size of ITIC (LA) could be regarded as 0 nm. For a suitable donor–acceptor blend system, one primary requirement is the formation of a continuous pathway with appropriate size for both donor and acceptor. However, in the S-TR:ITIC blend film, the aggregation of ITIC was inhibited, which is detrimental for electron transport. Considering the balanced charge transport, it is necessary to reduce the difference in the crystal size between S-TR and ITIC. As is well known, the use of CN as processing additives has emerged as a powerful approach to optimize crystallization in the bulk heterojunction blend system.[30−33] However, the crystallinity of both donor and acceptor in the reported blend systems always shows the same variation trend. Take poly(3-hexyl thiophene) (P3HT)/N,N-bis(1-ethylpropyl)-perylene-3,4,9,10-tetracarboxylic diimide (EP-PDI) blend system for instance, the solubility of EP-PDI and P3HT molecules in the CN solvent is much better than that in chlorobenzene (CB). Adding CN effectively facilitated the dispersion of EP-PDI molecules into the P3HT domains, thus inhibiting the crystallization process of both P3HT and EP-PDI.[30] However, in P3HT/[6,6]-phenyl-C61-butyric acid methyl ester blend system, the addition of CN could promote the formation of a highly crystalline structure of both donor and acceptor due to the higher vapor pressure (0.029 mm Hg).[33] Surprisingly, CN could have an opposite effect on the crystallinity of S-TR and ITIC in our work. It is obvious that the crystallinity of S-TR was decreased, whereas that of ITIC was enhanced after adding CN, as shown in Figure D–F. Furthermore, the gap in the crystal size between LD and LA was reduced, as shown in Table , with LD gradually reduced from 17.2 to 15.1 nm; nevertheless, LA increased from 0 to 12.9 nm, and we will discuss this phenomenon in the following paragraphs.

Table 1

Parameters of Crystal Size, Domain Size, Domain Purity, and Charge-Carrier Mobility of S-TR:ITIC Blend Films Processed Without and With Different Contents of CN

	crystal size (nm)
	S-TR	ITIC	domain size (nm)	domain purity (%)	μh (cm2 V–1 s–1)	μe (cm2 V–1 s–1)	μh/μe
W/O CN	17.2	0	57	100	2.36 × 10–3	1.13 × 10–5	208.9
1% CN	15.6	8.3			2.64 × 10–3	9.38 × 10–5	28.1
3% CN	15.1	12.9		69	1.35 × 10–3	3.82 × 10–4	3.53
5% CN	15.9	10.4			2.12 × 10–3	1.93 × 10–4	10.9

The crystallization of donor and/or acceptor governs the morphology in many blend systems.[34,35] Here, the morphology of S-TR:ITIC blend films was characterized by atomic force microscopy (AFM) and transmission electron microscopy (TEM). As shown in Figure S3, it is clear that S-TR is more prone to form grain-like crystals in neat S-TR film, and ITIC tends to form wire-like crystals in neat ITIC film. In S-TR:ITIC blend film (Figure A,a), only a grain-like morphology could be observed. With increase in the content of CN, the grain-like crystals gradually disappear. Meanwhile, more wire-like crystals begin to emerge, thus forming interpenetrating network (Figure C,c). When excessive CN exists, large crystals form and the interpenetrating structure is destroyed again (Figure D,d). The changed morphology should be attributed to the variation in the crystallinity of both S-TR and ITIC. If the crystallization of S-TR is dominant, a large and discontinuous phase separation structure would form due to the strong self-aggregation of S-TR. If the crystallization of S-TR is restrained and the crystallization of ITIC is promoted, the crystal size between S-TR and ITIC would be more balanced. Therefore, the competitive crystallization between S-TR and ITIC would promote the formation of an interpenetrating structure. Importantly, not only the morphology but also the carrier mobility indicates the formation of the complete network. By using the space-charge-limited current method, the charge-carrier mobilities can be well characterized (as shown in Figure S4), and the obtained data are summarized in Table .[36] For the blend system, the value of μh decreases to 1.35 × 10–3 cm2 V–1 s–1 from that of 2.36 × 10–3 cm2 V–1 s–1. In contrast, the mobility of electron in these thin films increases from 1.13 × 10–5 to 3.82 × 10–4 cm2 V–1 s–1. As a result, the lowest μh/μe ratio (3.53) occurs at a CN loading at 3%. The balanced hole and electron transport indicates the formation of an interpenetrating network, which should be ascribed to the reduced difference in the crystal size between the donor and the acceptor.[37]

Figure 3

Tapping-mode AFM height images and TEM images of S-TR:ITIC blend films processed without and with different contents of CN: (A, a) without CN, (B, b) with 1% CN, (C, c) with 3% CN, (D, d) with 5% CN.

Transmission resonant soft X-ray scattering (RSoXS) was used to study the phase separation as well.[29,38,39] To provide high materials contrast between S-TR and ITIC, we select the photon energy as 284.2 eV.[23] The average scattering sector profiles of the blend films processed without and with 3% CN are shown in Figure . The film without CN shows a peak at q = 0.11 nm–1. We estimated the length scale of phase separation (ξ) of those referential binary blends using ξ = 2π/qmode.[40] The length scale of 57 nm is found for the film fabricated without CN. Combined with the GIWAXS, it suggested that the phase separation in S-TR:ITIC blend system is mainly driven by the crystallinity of S-TR. However, for the film processed with CN, no clearly quantifiable peak could be observed in the measured q range, which indicates ξ is too small to be captured. It is believed that the competitive crystallization between S-TR and ITIC suppressed the appearance of a large-scale phase separation. In addition, the relative phase purity can be achieved by calculating the total scattering intensity through the integration of the scattering profiles over the q range and taking film thickness into account.[29,40,41] The blend film processed without CN shows a purity of 1, and it should be noted that the purity of S-TR:ITIC film without CN set to 100% here is only for comparison. Surprisingly, after adding CN, the phase purity of the blend film dropped to 0.69. Generally, it is suggested that a high average purity could reduce the bimolecular recombination, thus improved device performance was obtained. However, if the domain size is larger than the diffusion length of exciton, which means the domain purity is too pure (i.e., have insufficient nanomorphology), and the insufficient percolation pathways usually results in reduced device performance.[42,43] Apparently, in the S-TR:ITIC blend system, the crystal size of S-TR is too large, which induces too pure domain. After adding CN, the crystallinity of S-TR and ITIC was adjusted, thus obtaining an optimized purity, which leads to a trade-off between exciton diffusion and charge transport.

Figure 4

RSoXS profiles of S-TR:ITIC blend films processed without and with 3% CN.

Optimized Film-Forming Kinetics Leading to Balanced Crystal Sizes of Donor and Acceptor

A common approach to optimize BHJ morphology is to adjust the film-forming kinetics, which has profound impacts on the film structure, including phase separation, crystallinity, molecule orientation, and so on. Although lots of research has been conducted on film-forming kinetics, the process remains a “black art”, which has not been completely understood. The increased knowledge about the spin-coating process allows us to optimize the processing method more efficiently, providing a precise control over the phase-separation structure.[44−46] As we know, the in situ grazing incidence X-ray diffraction (GIXD), grazing incidence small-angle X-ray scattering, and optical experiments are widely applied to gain insight into the real-time structure evolution during the film-forming process.[44,45] Recently, Richter et al. developed an in situ photoluminescence (in situ PL). By evaluating the degree of PL quenching, we can monitor the phase-separation and crystallization process during the film-forming process.[46]

Here, we revealed the impacts of CN on the film-forming kinetics by tracking the PL intensity of both donor and acceptor during the spin-coating process. According to the PL spectra of pure S-TR and ITIC film (shown in Figure S2), the PL emission peak centered at 700 nm corresponds to the aggregation of S-TR and the signal around at 780 nm originates from aggregated ITIC. Hence, we recorded the change in these emission peaks (in situ PL as shown in Figure S5) to monitor the aggregated behavior of S-TR and ITIC during the spin-coating process. As shown in Figure , the PL spectra of the blend films processed without and with CN were measured. For the film processed without CN (Figure a), as the main solvent, i.e., CB, evaporates, the PL intensity of both S-TR and ITIC drops, consistent with a decreasing average distance between S-TR and ITIC (step I: about 0–9 s).[47] Then, both S-TR and ITIC show a sharp “dip” in the PL intensity followed by a recovery, termed here as step II (about 9–12 s). In this process, the recovery is attributed to the continuous evaporation of CB. The concentration of the solutes increases up to the saturated solubility, resulting in aggregates and phase separation of the solutes. In step III (about 12–24 s), the PL intensities of S-TR and ITIC reach plateaus, which is believed to be the increased phase purity balanced with the decreased distance between fluorophores and quenchers. In step IV, the PL intensity decreases rapidly first and then lowers slowly. We attributed the rapid decrease in early step IV (from 24 to 37 s) to enhanced crystallinity, which causes an increase in the exciton diffusion length, leading to an improved dissociation efficiency of exciton. However, in late step IV (37 s to the end), the morphology evolution ends with the complete evaporation of CB, and the gradual decrease in the PL intensity maybe attributed to the degradation of the film under strong illumination. It is distinct that the precipitation of S-TR and ITIC occurs at almost the same time at step II by analyzing the kinetics process. As a result, S-TR with a high crystallinity would inhibit the aggregation of ITIC to a certain extent, resulting in a small crystal size of ITIC.

Figure 5

PL spectra of S-TR:ITIC blend films processed without CN (A) and with 3% CN (B). The PL intensity (obtained from in situ PL spectra) vs time recorded during the spin-coating process (a) without CN and (b) with 3% CN.

In the blend system with CN, the change tendency of PL intensity is similar to that in blend system without CN during steps I and IV process as shown in Figure b. Surprisingly, it is different in steps II and III, especially regarding the PL intensity of ITIC. As we can see, in step II (about 10–19 s), as CB evaporates, S-TR forms aggregate, thus its PL intensity increases quickly and then reaches a maximum. However, only a portion of ITIC aggregates and forms crystal nucleus. Although most of ITIC disperses in the solvent, accordingly, the PL intensity of ITIC does not reach the maximum. In early step III (about 19–48 s), the PL intensity of ITIC remains constant, whereas in late step III (about 48–90 s), the PL intensity increased rapidly. This phenomenon can be attributed to the evaporation of CN, which promotes the dispersed ITIC to continue to aggregate.

Through a careful analysis, the working mechanism of CN for optimizing the morphology in the S-TR:ITIC blend system mainly has two aspects. First, CN is a selective processing additive, which has a lower solubility for S-TR (as shown in Table S1). As a result, S-TR would precipitate out first, i.e., the crystallization of S-TR and ITIC occurs at different stages, which reduces the interference of S-TR to the crystallization of ITIC. Second, the film-forming process was prolonged from about 30–200 s due to the lower volatilization rate, providing sufficient time for ITIC to self-organize. These two changes we mentioned above must be the reason for the increased ITIC crystal size after adding CN. In addition, it is found that adding CN reduces the S-TR aggregated degree in the solution (the PL peak of S-TR blueshifts after adding CN as shown in Figure S6), which may inhibit S-TR from nucleating during the film-forming process, leading to a decreased crystal size.

The study of the film-forming kinetics using in situ PL spectra is proposed for giving the selection criteria of additive to balance the crystal size of the donor and the acceptor in AS-NFSCs. First, the additive has a higher solubility for donor and acceptor than the main solvent. Second, the additive should have selective solubility for the component with weak crystallinity. Lastly, the volatilization rate of the additive should be slower than that of the main solvent.

Relationship between Morphology and Device Performance

To explore the influence of film morphology on the photoelectronic property, photovoltaic devices were prepared. The photovoltaic parameters of the S-TR:ITIC devices processed without and with different contents of CN are summarized in Table and the corresponding current density–voltage (J–V) curves and the external quantum efficiency (EQE) spectra are shown in Figure . The device parameters of the S-TR:ITIC BHJ layers as-cast shows a poor short-circuit current density (JSC) and fill factor (FF), with the power conversion efficiency (PCE) of only 1.75%. Fortunately, the S-TR:ITIC BHJ layers processed with CN exhibits a significant enhancement. After adding 3.0% CN, the corresponding device shows a PCE max of 3.59%, an improvement of about 2 times compared to that of the device without adding CN, with a JSC of 7.24 mA cm–2, a VOC of 1.00 V, and a FF of 0.50.

Table 2

Photovoltaic Parameters of S-TR:ITIC Blend System Processed without and with Different Contents of CN

	VOC (V)	JSC (mA cm–2)	FF	PCE (%)	JSC:S-TR (mA cm–2)	JSC:ITIC (mA cm–2)
W/O CN	0.99 ± 0.01	4.41±0.12	0.40 ± 0.04	1.75 ± 0.15	3.08	1.36
1% CN	0.99 ± 0.01	5.20 ± 0.23	0.45 ± 0.01	2.36 ± 0.36	3.31	1.73
3% CN	1.00 ± 0.01	7.24 ± 0.21	0.50 ± 0.02	3.59 ± 0.18	4.64	2.49
5% CN	0.97 ± 0.02	6.30 ± 0.17	0.41±0.02	2.53 ± 0.21	4.10	1.95

Figure 6

J–V curves (A), EQE curves (B), and (C) JSC–P curves of photovoltaic devices for S-TR:ITIC blend system processed without and with different contents of CN.

The EQE curves are shown in Figure B. The relative magnitudes of the EQE spectra correlate well with the JSC values determined from the J–V curves.[48] Although the addition of CN promotes a blueshift in the UV–vis spectra of S-TR, a prominent improvement can be observed in the wavelength ranging from 350 to 780 nm after adding CN. Quantitative analysis of this effect has been performed by integrating the product of EQE, in which the spectra in the range of 320–620 and 620–780 nm refer to these quantities as JSC:S-TR and JSC:ITIC, respectively (Table ). It is found that both JSC:S-TR and JSC:ITIC exhibit a great increase after adding 3.0% CN. The increased JSC:S-TR must be partly caused by the increased exciton dissociation efficiency due to the blueshifted absorption spectra and the decreased crystallinity of S-TR. It should be noted that the crystal size correlates with the smallest pure domain size, which should match with the typical exciton diffusion length. As we mentioned above, LD was reduced from 16.4 to 14.5 nm after adding CN, promoting the exciton generated in the S-TR phase to diffuse to the interface and dissociate. The increased JSC:ITIC may mainly results from the increased crystal size of ITIC, which promoted the formation of the electron pathways and enhanced the electron mobility.

The other important reason for the improvement of both JSC:S-TR and JSC:ITIC is the suppressed bimolecular recombination. The light-intensity dependencies of the device performances were studied as well to identify the charge recombination origin. The relationship between JSC and light intensity follows the relation JSC ∝ P (P is the light intensity and S is the exponential factor). If there is no bimolecular recombination, S factor should be 1, whereas if the bimolecular charge recombination occurs, S is smaller than 1. As shown in Figure C, the data are plotted on a log–log scale, and S in the devices were calculated as 0.89, 0.92, 0.96, and 0.95 for the films processed without CN and with different contents CN, respectively. This indicates that the bimolecular recombination was minimized after adding CN. Without the addition of CN, LA is too small and no continuous pathway is formed for electron transport. The mismatch in crystal size between S-TR and ITIC could lead to an imbalance in the carrier transport (μh/μe = 208.9). After adding CN, the crystal size between the donor and the acceptor became more symmetric, forming an interpenetrating network. As a result, a balanced charge transport was obtained (μh/μe = 3.53). Balancing the charge mobility between the hole and the electron is an effective way to reduce bimolecular recombination, which also agrees well with the remarkable increase in FF. We also analyze VOC as a function of light intensity, as shown in Figure S7. It is obvious that the slope for the device with CN (1.59kT/q) is smaller than that of the device without CN (1.87kT/q), suggesting that the optimized morphology could effectively reduce the interfacial trap-assisted recombination and accordingly result in enhanced device performance.[49]

Conclusions

We have demonstrated that balancing the crystal size between the donor and the acceptor is a crucial premise for fabricating the interpenetrating network in the S-TR:ITIC blend system. The crystal size of S-TR was decreased and that of ITIC was increased, which should be attributed to the optimized solution state and film-forming kinetics. The nucleation barrier of S-TR in a solution was increased and the disturbance of S-TR on the self-organization of ITIC during the film-forming process was reduced. The regulated crystal size can effectively promote the formation of the interpenetrating network, which facilitated the exciton dissociation and suppressed the bimolecular recombination, resulting in an improved PCE. Additionally, three criteria for processing additive for reducing the gap of crystal size between the donor and the acceptor have been identified: (i) additive has a higher solubility for the donor and the acceptor than the main solvent; (ii) additive should have selective solubility for the component with weak crystallinity; (iii) the volatilization rate of the additive should be slower than that of the main solvent. Our work demonstrated the precondition for fabricating the interpenetrating network in AS-NFSCs and provided the approach for regulating the crystal size by using additive strategy.

Experimental Section

Materials

BDTT-S-TR (S-TR) and ITIC were purchased from Solarmer Materials Inc. Anhydrous solvents chlorobenzene (CB) and chloronaphthalene (CN) were purchased from Sigma-Aldrich. All of the chemicals were used without further treatment.

PSCs Fabrication

Organic solar cells were fabricated in the following configuration: indium tin oxide (ITO)/S-TR:ITIC/MoO3/Al. We used UV–ozone to treat the cleaned ITO glass substrates for 25 min and then spin-coated ZnO layer on the top of the ITO glass. Subsequently, the ZnO layer was dried at 200 °C for 60 min. The S-TR:ITIC solution containing different contents of CN with a total concentration of 12 mg mL–1 was spin-cast on top of the ZnO layer and the thickness of the active layer was about 140 nm (the solution temperature was 100 °C). Finally, MoO3 (10 nm) layer and Al (100 nm) layer were deposited on top of the active layer in turn by thermal evaporation in a vacuum of 4 × 10–4 Pa. Finally, four solar cells were fabricated per ITO glass substrate, each with an active area of 7.2 mm2.

Characterization

The absorption of the sample was recorded by UV–vis absorption spectroscopy (AvaLight-Hal) with a halogen lamp source. The fluorescence spectra of the pure and the blend films were recorded by spectrometer (Jobin Yvon LabRAM HR spectrometer with a 380 nm solid state laser excitation light source). The in situ fluorescence spectra were recorded using a homemade instrument, which contained a spin-coater (WE!NVIEW) and a spectrometer (Ocean USB2000+).

The morphology of the S-TR:ITIC film was characterized by using atomic force microscopy (AFM) and transmission electron microscopy (TEM). The AFM was employed to study the surface topography of the films. The images were recorded by the combined use of SPI3800N AFM (Seiko Instruments Inc., Japan) with a Si tip (the spring constant is 3 N m–1). In our work, the AFM was performed in the tapping mode. The TEM experiments were performed by using a TEM-1011 (JEOL Co., Japan), and the accelerating voltage is 100 kV.

The phase purity was analyzed by a RSoXS transmission measurement, which was performed at beamline 11.0.1.2 at the advanced light source (ALS). The films for RSoXS measurement were prepared on a PSS-modified Si substrate, and the fabrication process is the same as those used for device fabrication. Then, the film was transferred by floating in water to a Si3N4 membrane (Norcada Inc.). The two-dimensional scattering patterns were recorded on an in-vacuum CCD camera (Princeton Instrument PI-MTE). The sample detector distance was calibrated from the diffraction peaks of poly(isoprene-b-styrene-b-2-vinyl pyridine), and it had a known spacing of 391 Å. The beam size at the sample was 100 × 200 μm2.

The crystallinity of the blend film was analyzed using a grazing incidence wide-angle X-ray scattering (GIWAXS), which was performed at the beamline 7.3.3 advanced light source (ALS; Lawrence Berkeley National Laboratory). The films were prepared on a Si substrate, and the fabrication process is the same as that used for device fabrication. The 10 keV X-ray beam was incident at a grazing angle of 0.12–0.16°, selected to maximize the scattering intensity from the samples. A photon-counting detector (Dectris Pilatus 2M) was employed to detect the scattered X-rays, and used silver behenate (AgB) to calibrate.

We use a computer-controlled Keithley 236 source to measure the current density–voltage (I–V) characteristics under AM1.5G illumination from a calibrated solar simulator, and the irradiation intensity is 100 mW cm–2. We use a lock-in amplifier at a chopping frequency of 280 Hz during illumination with a monochromatic light from a xenon lamp to measure the external quantum efficiency (EQE).