Seon-Mi
Jin
a,
Jun Ho
Hwang
a,
Jaehyeong
Park
a,
Du Hyeon
Ryu
b,
BongSoo
Kim
c,
Chang Eun
Song
b and
Eunji
Lee
*a
aSchool of Materials Science and Engineering, Gwangju Institute of Science and Technology (GIST), Gwangju 61005, Republic of Korea. E-mail: eunjilee@gist.ac.kr
bEnergy Materials Research Center, Korea Research Institute of Chemical Technology (KRICT), Daejeon 34114, Republic of Korea
cDepartment of Chemistry, Ulsan National Institute of Science and Technology (UNIST), Ulsan 44919, Republic of Korea
First published on 25th October 2023
Achieving precise control over the 3D morphology of the active layer in all-polymer solar cells (APSCs) is crucial for improving power conversion efficiency and stability. The ideal configuration involves a vertically aligned p–n nanoheterojunction with a 10–20 nm exciton diffusion length, but the inherent freedom of polymer chains poses challenges due to their stochastic nature. Extensive exploration of spatial analytical techniques, from molecular to macroscale levels, is also needed to reveal the polymer behavior governing the active layers. Meanwhile, halogenated solvents are used for high-efficiency APSC fabrication due to their favorable solubility with each polymer component. However, recent efforts aim to harness non-halogenated solvents, mitigating toxicity and environmental hazards. Here, we explore the generation of uniform thin film morphologies based on solution-processable crystallization-driven polymer assemblies. Pre-assembled n-type crystalline nanowires (NWs) via heating and cooling solutions of non-halogenated 1,2,4-trimethylbenzene formed very well-aligned, uniform thin film structures with well-dissolved noncrystalline p-type polymers and showed efficiency comparable to the performance of thin films prepared with chlorobenzene (CB) halogenated solvent. X-ray scattering and transmission electron microtomography confirm organized, long-range aligned one-dimensional NWs with closely interfaced p–n nanoheterojunctions through strong intra- and inter-polymeric π-stacking interactions. Notably, the resulting thin film morphology resembles that of blended film casting with CB. Significantly, the long-term stability of the crystalline NWs-based thin film morphology under light exposure surpasses that of blend films due to robust intermolecular packing in the NWs.
Charge transport in organic semiconductors is a complex process ranging from molecular level to macroscopic scales (typical film thickness is approximately 100 nm),15 which is due to the semicrystalline nature of conjugated polymers. Understanding the charge transfer process at each scale is critical to improving the performance of electronic devices, such as organic field effect transistors16,17 and organic solar cells.18,19 At the molecular level, the torsion angle between monomer units in the polymer chain governs the degree of electronic coupling and chain coplanarity.20 Untwisted chains promote efficient charge transfer within the chain, while the arrangement of molecules in aggregates or crystallites influences charge hopping along the chain. Favorable alignment and overlap of π-electron clouds between adjacent molecules, known as good π–π stacking, create more opportunities for effective charge movement along the chain.21
The active layer of APSCs is comprised of donor and acceptor polymers, and the morphology of the active layer plays a crucial role in the device performance.22,23 One of the most critical factors determining the efficiency of APSCs is achieving proper phase-separated donor–acceptor domain dimensions, as the exciton diffusion length is limited to approximately 10–20 nm.24 In addition, the connectivity of domains plays an important role in regulating charge flow to control the mobility of charge carriers to the corresponding electrodes, which is essential to minimize carrier recombination. The formation of aggregates and crystalline domains, along with the orientation of the π-conjugated backbone with respect to the electrode surface, significantly influences charge flow and its direction.25 Molecular characteristics (molecular weight, molecular weight distribution, regioregularity, backbone, and side-chain structure, etc.),26–28 solvent additives, and post-deposition treatments (solvent vapor annealing, thermal annealing, etc.)29–31 can impact the PCE of APSCs. During solution casting, it is challenging to construct a bicontinuous network structure due to various polymer interactions that may result in gradient differences of components along the longitudinal direction of the film.32 Therefore, microscopic morphological constraints enable efficient charge separation and the development of charge transport channels with limited bimolecular recombination. Another critical issue for achieving optimal charge generation and collection in APSCs relies on controlling the interfacial orientation and molecular packing of the long anisotropic conjugated polymer chains in the composites.33–35 However, non-optimized molecular packing and interfacial orientation can reduce charge collection efficiency, so it is important to develop precise strategies to control polymer packing and orientation to realize high-efficiency APSCs.
Recent breakthroughs in transmission electron microscopy (TEM) have enabled the characterization of structural details with nanometer resolution in three dimensions (3D), leading to the provision of a reconstructed volume image based on computed tomography. This advanced imaging technique has been acknowledged as an efficacious tool in unraveling the morphology of polymers: fullerene and organic–inorganic hybrid blend films.36,37 Our recent research demonstrated the significant impact of the 3D morphologies of APSC blend films on device efficiency, using TEM tomography (TEMT) analysis.38 The long-term stability against light irradiation was also explained with morphological degradation.36,38 Thus, 3D visualization of blend films containing donor and acceptor materials with different optoelectrical properties can provide a better understanding of film morphologies and offer important strategies for improving the device performance and stability.
In terms of a more environmentally friendly approach to solution processing of APSCs, it is important to consider the use of non-toxic, low-volatile green solvents and additives.39 Current standards for the most efficient APSCs still rely heavily on the utilization of halogenated solvents and additives to achieve optimal solubility and active blend morphology. However, given the detrimental effects of halogenated solvents on both human health and the environment, a shift towards non-halogenated solvents is of paramount importance.40,41 Recent studies have demonstrated promising photovoltaic properties in polymer blend films processed using non-halogenated solvents. For their large-area module production, appropriate solvents that can induce uniform coating in roll-to-roll processing need to be explored.
In this study, we aimed to establish a comprehensive correlation among the intricate interplay of structure, and morphology of active layer with performance in APSCs. To achieve this, we opted for the non-halogenated solvent, 1,2,4-trimethylbenzene (TMB), as an alternative to traditional halogenated solvents. This allowed us to precisely control the distribution of the components, molecular arrangement, and direction in the poly[4,8-bis(5-(2-ethylhexyl)thiophen-2-yl)benzo[1,2-b;4,5-b′]dithiophene-2,6-diyl-alt-(4-(2-ethylhexyl)-3-fluorothieno[3,4-b]thiophene-)-2-carboxylate-2-6-diyl] (PTB7-Th) and poly{[N,N′-bis(2-octyldodecyl) naphthalene-1,4,5,8-bis(dicarboximide)-2,6-diyl]-alt-5,5′-(2,2′-bithiophene)} (P(NDI2OD-T2)) blend films. In the TMB solvent, the highly crystalline n-type P(NDI2OD-T2) formed 1D nanostructures through strong intermolecular interactions. These nanostructures were subsequently combined with the p-type PTB7-Th to create a highly uniform bulk heterojunction (BHJ) active layer. This layer was arranged alternately over a large area through induced assembly, both vertically and horizontally, along with the orientation of the P(NDI2OD-T2) chains. The 3D morphology and uniformity of the active layer were effectively characterized using TEMT, allowing qualitative and quantitative visualization. The well-organized BHJ nanolanes facilitated efficient charge transport between the crystals, resulting in photovoltaic performance surpassing that of halogen solvents. Furthermore, the photovoltaic performance was further enhanced through a straightforward thermal annealing process. This process significantly improved the short-circuit current density (JSC) and fill factor (FF). The strong intermolecular interactions within the active layer, formed by crystalline 1D nanostructures, contributed to its relatively high photostability. These findings offer valuable insights and strategies for optimizing the active layer using a non-halogenated solvent at various scales. Ultimately, this research deepens our understanding of the complex structure–property correlations within APSCs.
To determine whether nonhalogenated TMB with relatively large Ra compared to other halogenated solvents induces self-aggregation of PTB7-Th and P(NDI2OD-T2), we observed a polymer (0.1 mg mL−1) using TEM. The neat film as-cast from PTB7-Th TMB solution did not show any noticeable structures (Fig. 1a and S2a†). However, P(NDI2OD-T2) neat film shows mainly leaf-like network structures composed of bundles of fibrils (hereafter referred to as NW, Fig. 1b and S2b†). Polarized optical microscopy (POM) image revealed that the spontaneously formed NWs were birefringent, indicating the presence of polymer crystallites (inset of Fig. 1b). The PTB7-Th TMB solution does not show any birefringence (inset of Fig. 1a). Interestingly, when a film was manufactured after mixing two polymer-TMB solutions, well-aligned NW arrays in a large area were confirmed (Fig. 1c and S2c†). The P(NDI2OD-T2)-rich phase displayed a lighter appearance compared to the PTB7-Th-rich phase, consistent with the observations in a previous study.38 Note that the film of the solution mixed in CN with a low Ra value does not have distinct phase separation (Fig. 1d).
In order to obtain detailed spatial information about the lateral and vertical morphology and each domain sizes of each component in NW-based TMB films of PTB7-Th/P(NDI2OD-T2), TEMT was conducted (Fig. 1e, S3 and Movie S1†).45,46 The volume composition of the two polymers could be investigated through weighted back-projection algorithm-based reconstruction of a series of 2D projections obtained at 2° degree intervals from −68° to 70° degrees relative to the electron beam incident angle (Table S3†). Since the reconstructed images showed stronger contrast than the conventional TEM image due to the utilization of a filtered algorithm, the nanofibrillar morphology could be observed more clearly in the 3D volume rendering image (Fig. 1e and f). Indeed, P(NDI2OD-T2)-rich domain appeared in a lighter phase than the PTB7-Th-rich domain in the bright-field TEM images.38 In the 3D volume image, the bright and dark phases were marked with P(NDI2OD-T2) in yellow and PTB7-Th in sky blue, respectively (Fig. S3†). When xy slices of the entire tomogram were extracted, we found that the unidirectional growth 1D nanostructures of the BHJ film were aligned over a large area (Fig. 1f). The size of P(NDI2OD-T2) domain was estimated to be the 9–10 nm wide and several micrometers long. Unlike conventional TEM, TEMT clearly further showed the vertical phase separation in the entire volume of a film.36–38,47 The vertical (xz) image showed good interconnectivity of each polymer (Fig. 1g). The 3D morphology of the PTB7-Th/P(NDI2OD-T2) photoactive layer that satisfies the exciton diffusion length of 10–20 nm and has a percolation structure connecting both electrodes affects charge generation, charge recombination inhibition, and efficient charge transfer, making it very reasonable for APSC application. The relevant results will be explained later.
The UV-Vis absorption spectra of each P(NDI2OD-T2) and PTB7-Th/P(NDI2OD-T2) blended solutions were investigated to determine whether the P(NDI2OD-T2) NWs were precrystallized in TMB and transferred its characteristic into the film to form such morphology (Fig. 1h).48 We observed two absorption bands of P(NDI2OD-T2) TMB solution, a high-energy peak at a wavelength of approximately 400 nm attributable to the π–π* transition and a low-energy band (500–900 nm) attributed to intramolecular charge-transfer transition, implying that crystallization-induced aggregation between polymer chains occurred in the solution-state.48 The absorption peak of P(NDI2OD-T2) TMB solution reached to red-to-NIR region, with a peak maximum at 705 nm and a shoulder at ∼820 nm, compared to CN solution, which lacked a low energy absorption peak and showed a broad, featureless band with a peak maximum at 615 nm. No aggregates were observed in the TEM image of the CN solution, as mentioned earlier (Fig. 1d). In a blending TMB solution containing PTB7-Th and P(NDI2OD-T2) aggregates (hereafter referred as to 1Pre), four major peaks were observed at approximately 330, 390, 640, and 705 nm. The peaks at 330 and 640 nm were attributed to the PTB7-Th, the peak at 390 nm was attributed to the absorption of the P(NDI2OD-T2), and the peak at 705 nm was attributed to both the PTB7-Th and P(NDI2OD-T2). The characteristic absorption peaks of the P(NDI2OD-T2) aggregates formed via solution-state crystallization-driven assembly were observed to be well-maintained in the blending solution with the PTB7-Th polymer, which is supported by the POM image showing birefringent (Fig. 1c). The grazing-incidence wide-angle X-ray scattering (GIWAXS) was performed to investigate the crystalline order of P(NDI2OD-T2) affecting the charge transport (Fig. 1i–k and S4, Table S4†). The thin film as-cast from a 1Pre solution exhibited the crystalline characteristic (100) peak with a d-spacing of 26.3 Å and (010) peak with a d-spacing of 3.95 Å along out-of-plane (OOP) direction (qz). Note that the pristine PTB7-Th film is simply amorphous, displaying the face-on molecular orientation with a strong intermolecular π–π stacking (010) peak of PTB7-Th in OOP, while the pristine P(NDI2OD-T2) film is crystalline, showing a lamellar stacking (100) peak of P(NDI2OD-T2) in OOP (Fig. S4a and b†). These results suggested that the crystalline character of P(NDI2OD-T2) induced in TMB is well-preserved in the cast film.
The photovoltaic properties of APSC using TMB-processable NWs-based thin film of 1Pre as an active layer were investigated by manufacturing an inverted device structure of indium tin oxide (ITO)/ZnO/PEIE/PTB7-Th:P(NDI2OD-T2) photoactive layer/MoOx/Ag. The photovoltaic characteristics of the PTB7-Th/P(NDI2OD-T2) APSCs are displayed in Fig. 2 and summarized in Table 1. The PCE of the thin film fabricated using 1Pre was measured to be 4.84%, while the PCE of a film cast by dissolving two polymers in the halogenated solvent CB (denoted to 2) was measured to be 5.68%. Interestingly, a device with solution of two polymers dissolved in TMB at once (denoted 1Post) applied to the active layer showed a PCE of only 2.79% (Scheme S2†). Considering that in-plane (IP) and OOP crystal coherence lengths (CLs) of the thin film prepared by 1Pre and 1Post are 161.3 Å and 25.4 Å and 147.7 Å and 19.4 Å, respectively, estimated using Scherrer equation49,50 (Table S4†), it is conceivable that increasing the crystallinity of the p-type or n-type polymers inside the films could further improve the PCE. In a 1Post solution, the crystallization of P(NDI2OD-T2) can be hindered because they self-assemble in the presence of PTB7-Th. Along this line, the 1Pre-H solution was prepared by heating and cooling the P(NDI2OD-T2) TMB solution to make it sufficiently crystalline in the solution and mixed with the PTB7-Th solution. The resulting thin film was heat-treated to obtain a 1-T thin film. The morphology of 1Pre-H and 1-T thin-films showed similar NW-based structures to that of 1Pre (Fig. S6†). However, the device performance with 1Pre-H and 1-T thin-films showed excellent improvement as expected, with average PCEs reaching 5.89% and 6.12%, respectively (Fig. 2 and Table 1). These values exceeded the PCE of the thin film cast from 2 using CB, which are achieved from the high JSCs (>10.0 mA cm−2) and increase in the FFs (>0.50). After thermal treatment of precrystalline NW-based 1-T thin film (prepared by 1Pre-H), the crystallinity of the P(NDI2OD-T2) nanodomain improved and the electron mobility of the whole device increased, which ultimately led to a significant improvement in JSCs and FFs. The film made of 1Post showed the lowest PCE (2.79%), which is consistent with the low open-circuit voltage (VOC), JSC, and FF values, which is a result of the low crystallinity of P(NDI2OD-T2) originating from the non-appropriate solution-processing. The lowest external quantum efficiency (EQE) was estimated, with a maximum value of ∼55%.
Experimenta | Method | Processing solvent | V OC [V] | J SC [mA cm−2] | FF [%] | PCE [%] |
---|---|---|---|---|---|---|
a Inverted device architecture is ITO/ZnO NPs/PEIE/photoactive layer (d = ∼100 nm)/MoOX/Ag. b J SC values calculated from the EQE spectra. c The values in parenthesis are average photovoltaic properties obtained from over 10 devices. | ||||||
1 Pre | Pre-crystalline P(NDI2OD-T2) NWs mixed with PTB7-Th | TMB | 0.79 (0.78 ± 0.01)c | 12.88 (12.38)b (12.68 ± 0.21)c | 47 (45 ± 2)c | 4.84 (4.58 ± 0.27)c |
1 Pre-H | Pre-crystalline P(NDI2OD-T2) NWs (by heating and cooling) mixed with PTB7-Th | TMB | 0.83 (0.82 ± 0.01)c | 13.28 (12.75)b (13.11 ± 0.16)c | 53 (52 ± 1)c | 5.89 (5.74 ± 0.16)c |
1-T | Thermal annealing of cast film 1Pre-H | TMB | 0.84 (0.83 ± 0.01)c | 13.59 (13.06)b (13.45 ± 0.15)c | 53 (52 ± 1)c | 6.12 (5.96 ± 0.17)c |
1 Post | Crystallization of P(NDI2OD-T2) NW in PTB7-Th | TMB | 0.63 (0.60 ± 0.02)c | 11.92 (11.50)b (11.67 ± 0.25)c | 36 (33 ± 2)c | 2.79 (2.43 ± 0.35)c |
2 | PTB7-Th and P(NDI2OD-T2) Blends | CB | 0.82 (0.81 ± 0.01)c | 13.20 (12.68)b (13.02 ± 0.17)c | 52 (52 ± 1)c | 5.68 (5.45 ± 0.24)c |
After the dissociation of excitons into free charge carriers, BHJ photoactive films with higher and balanced carrier mobility generally show high JSC and FF values, which usually originate from the well-ordered molecular packing in APSCs. We used the space-charge limited current (SCLC) method to measure the hole mobilities (μh) and electron mobilities (μe) of PTB7-Th/P(NDI2OD-T2) blend films (Fig. 2d, S7† and Table 2). Both 1Pre-H and 1-T films could enhance μe as well as μh. The mobility reached a peak in the 1-T film, with μh of 1.86 × 10−4 cm2 V−1 s−1 and μe of 5.79 × 10−5 cm2 V−1s−1. The ratio of hole and electron mobility (μh/μe) showed a value of 7.82 in the film of 1Post and gradually balanced to 3.21 in the film of 1-T. The thin film of 1Post showed the lowest μh and μe.
Experiment |
μ
h![]() |
μ
e![]() |
μ h/μe |
---|---|---|---|
a Hole-only device is ITO/PEDOT:PSS/photoactive film (d = ∼100 nm)/Au. b Electron-only device is ITO/ZnO NPs/PEIE/photoactive film (d = ∼100 nm)/Ca/Al. | |||
1 Pre | 0.84 × 10−4 | 1.50 × 10−5 | 5.60 |
1 Pre-H | 1.59 × 10−4 | 4.91 × 10−5 | 3.24 |
1-T | 1.86 × 10−4 | 5.79 × 10−5 | 3.21 |
1 Post | 0.38 × 10−4 | 0.49 × 10−5 | 7.82 |
2 | 1.08 × 10−4 | 3.20 × 10−5 | 3.38 |
The polymer orientation within the nanoscopic bicontinuous interpenetrating networks consisting of p–n nanojunctions with well-controlled polymer domains is critical to facilitate charge carrier transport via controlled interchain interactions within each 1D P(NDI2OD-T2) and PTB7-Th domains. The blend films had face-on structures with the pronounced (010) peak in the OOP direction (Fig. 1k), as mentioned earlier. To further corroborate the well-controlled morphology toward the corresponding interlayers and electrodes within the photoactive layer with constant thickness, the composition of each polymer at the top and bottom of the film was quantitatively investigated by area calculation from binarized xy-slice images (Fig. 3a). The binarized images were extracted from TEMT. The P(NDI2OD-T2) content was found to be relatively uniform, 50–55%, regardless of the top and bottom of the photoactive layer. To more quantitatively describe the orientation of the p–n nanoheterojunction arranged in the direction of the electrodes with long-range order, the orientation was color-mapped using the GTFiber software,51 an automated image analysis (Fig. 3b and S8†). The average 1D NW arranged along the long axis of the fibrous structure with the dominant orientation was colored green, while NWs that deviated from this alignment were labelled in a reddish tint. In addition, the orientation of aligned NW structures in each xy-slice was quantified via orientation parameter (S2D, Fig. 3c).51 An S2D of 1 means that the NWs are perfectly aligned in the same direction, while an S2D closed to 0 means that the NWs are randomly aligned. In all xy-slices, S2D values in the range of 0.87 to 0.90 by determination show that the NWs are regularly aligned in the entire volume. This morphology led to improved photovoltaic properties and device performance.52,53 This claims are also supported by the PTB7-Th/P(NDI2OD-T2) blend morphology generated in halogenated solvents, CB.
![]() | ||
Fig. 3 (a) XY-slice images from bottom to top (left to right) extracted from 3D construction (as depicted in Fig. S3†) of PTB7-Th/P(NDI2OD-T2) NW thin film, cast from the 1Pre solution. The PTB7-Th- and P(NDI2OD-T2)-rich domains are represented in black and gray, respectively. (b) Corresponding color-mapping and (c) pole figures showing the orientation distribution of P(NDI2OD-T2) NWs. |
The morphology of each polymer was examined in the CB solution, which showed a relatively high PCE value without any additives (Fig. 4a and b). Surprisingly, the TEM image of the P(NDI2OD-T2) polymer solution showed the formation of well-ordered 1D NWs. The CB solution of PTB7-Th did not show any noticeable aggregates. We further found that these fibrillar structures formed over a very large area in the PTB7-Th/P(NDI2OD-T2) blend thin film (Fig. 4c), similar to what was observed in the TMB solution. Comparison of GIWAXS data of thin films cast with 1Pre-H and 2 solutions showed similar results. However, it is known that increasing the proportion of face-oriented crystallites can improve the electron mobility and μh/μe balance of PTB7-Th/P(NDI2OD-T2), so we extracted the pole figure of the (100) lamellar crystalline peak was constructed by extracting the integrated intensity as a function of the azimuth angles (χ) along the arcs (Fig. 4e).54 The relative portions of edge-on and face-on orientation of crystallites were determined by the integral values of each section in the pole figure. The χ ranges of 0–45° and 45–90° relative to the surface normal are defined as edge-on and face-on orientations, respectively. The incorporation of PTB7-Th slightly increased the proportion of integrated area at 45–90°, showing an increase in the proportion of face-on oriented chains. This trend was stronger in the thin films cast with the CB solution. Therefore, identifying the morphology of the most efficient photoactive layer based on polymer material design and achieving the optimal morphology through molecular modification is an important strategy to improve APSC device performance.
To further investigate the device photostability of the NW-based thin film in response to sunlight, APSCs encapsulated with cover glass were exposed to continuous illumination from an AM 1.5 G solar simulator. The evolution of the current density–voltage (J–V) and photovoltaic characteristics are represented over time (Fig. 4g). Device from 1Pre showed a considerable burn-in loss (50% reduction of its initial PCE) during the first 50 h. The severe degradations of JSC and FF were the main factors influencing the extreme burn-in loss in efficiency. On the contrary, long-term photostability with relatively low burn-in loss was observed in APSCs of 1Pre-H and 1-T. Which retained approximately 80% of their initial PCE for up to 300 h. Approximately 80% of the initial PCE was observed to be maintained for up to 300 h. For the device of 2, the initial efficiency was similar to 1Pre-H and 1-T, but the degradation was relatively large, especially for 1-T, which can be explained by the improved intra-, intermolecular π–π interactions and crystallinity of the polymer after thermal treatment, which increases the resistance to photooxidation. Among the degraded photovoltaic parameters, the decrease in JSC and FF is closely related to the decrease in charge carrier mobility and increase in trapped states due to BHJ morphology degradation. As a result, we can demonstrate that the crystalline NW-based photoactive films prevent morphological degradation than conventional photoactive film, affecting the long-term stability of devices.
![]() | (1) |
Footnote |
† Electronic supplementary information (ESI) available: Experimental procedures, polymer characterisation, additional spectroscopy and results, and device characteristics. See DOI: https://doi.org/10.1039/d3py00718a |
This journal is © The Royal Society of Chemistry 2023 |