Introduction

Achieving long-term stability is both critical and intricate in the progression of organic solar cells (OSCs), requiring thorough scrutiny and innovative solutions1,2,3,4,5,6,7. Given their multilayered structure, OSCs intrinsically grapple with stability challenges. These difficulties are intensified when exposed to a myriad of stress factors such as irradiation, oxygen, moisture, thermal conditions, and mechanical strains during device operation8,9,10,11,12. In general, to understand the degradation mechanisms of OSCs, one must take into account various facets, including (1) the transition of aggregates or crystalline structures and phase separations from the thermodynamic disequilibrium to the equilibrium case13,14,15,16,17; (2) the mechanical integrity is compromised owing to diffusion-induced cohesive failures, micro-delamination, and the impacts of defect states12,18,19,20,21; and (3) functional impairments encompass chain free-radical oxidation reactions, bond severance and crosslinking, modifications in energy levels, and photoisomerization events22,23,24,25,26,27. In addition to the aforementioned factors, an often underemphasized yet crucial element could be the significance of the electron transport network in dictating device stability. It becomes evident that the sustained, efficient performance of OSCs hinges on the electron transport network’s consistency during extended device operation.

Particularly, the emergence of non-fullerene acceptors (NFAs) prompts questions regarding their capability to establish a robust electron transport network essential for stable OSC devices. Notably, fullerene derivatives are advantageous in forming a stable three-dimensional (3D) electron transport network, given the nearly spherical molecular configuration of fullerene molecules28,29. On the other hand, the coplanar molecular configurations of cutting-edge NFAs promote the π-π stacking of NFA molecules, which can also inherently support the formation of a 3D electron transport network in such bulk heterojunction (BHJ) films30,31. However, the connectivity of the electron transport framework of NFAs strongly relies on the molecular orientation and the intermolecular interaction, leading to a fragile and unstable network, which is more susceptible to degradation or localized interruptions in electron transport pathways.

On the other hand, it is crucial to point out the charge transport mechanism of polycrystalline organic semiconductors. Firstly, the fastest charge transport occurs along the backbone of conjugated polymers, due to the high intra-chain electronic coupling along the conjugated direction. Subsequently, the next favored transport orientation is in alignment with the π-π stacking observed between π-conjugated organic molecules or polymers32,33. In addition, organic semiconductor films commonly display a polycrystalline microstructure, with smaller crystalline regions separated by grain boundaries. In such a polycrystalline morphology, the charge transport within the crystalline domains can be highly efficient, while the bottleneck for charge transport could be at the grain boundaries34,35. Consequently, organic semiconductors that exhibit enhanced connectivity between domains are anticipated to provide superior charge transport performance. Given this context, polymer acceptors might possess a distinct advantage over their small molecular counterparts in terms of charge transport due to their potentially superior inter-domain connectivity. As devices aging and degradation, defect state effects and alteration of molecular conformation can significantly influence the electron transport properties of BHJ films. In these scenarios, a more cohesively interconnected charge transport framework would be advantageous, offering resistance and robustness against the scattered failures of individual transport units.

In this contribution, we discovered that the connectivity of the electron transport network is a key factor influencing the stability of OSCs. Our investigation focused on analyzing the charge carrier transport properties of active layers based on different acceptors, including small molecules, polymers, and oligomers. The results show that the D/A blends based on polymeric acceptors exhibit lower percolation thresholds, better tolerance to impurities, and less degradation in electron transport properties than those based on oligomers and small-molecule acceptors. The superior electron transport property of polymer acceptors in D/A blends can be attributed to the better connectivity of polymer units that enables a more robust network, resisting local degradation and intrusion of impurities. Furthermore, we compared the stability of devices and morphologies based on PY-V-γ and Y6 BHJs, which have nearly identical core structures except for the number of monomer units. The results indicate that the device stability of OSCs is influenced by electron transport network interconnectivity, which originates from a smaller diffusion coefficient and mixing entropy of polymeric acceptors.

By employing the extended Su-Schrieffer-Heeger (SSH) tight-binding model, we observed that the establishment of the connectivity requires an electron transfer integral larger than 0.05 eV. Moreover, this connectivity remained stable even with increasing intermolecular distances. Meanwhile, random electron passing tests and the algebraic connectivity analysis further confirm the desirable connectivity of the electron transport network contributing to effective electron transmission and device stability in such polymeric conformations. Overall, this work provides an important material design principle from the perspective of charge carrier transport to fabricate stable OSCs.

Results

Impact of polystyrene content and donor:acceptor composition variations on electron transport

In OSCs, the inherent fragility of organic semiconductor films makes them sensitive to factors beyond optimal preparation conditions, such as the introduction of impurities and variations in D:A ratios. These factors directly influence the interpenetrating networks within the active layer, thereby affecting exciton dissociation, charge generation, charge transport processes, and exacerbating aging. To investigate the stability of organic semiconductor devices from the perspective of charge transport, we employ various D/A ratios and impurity (polystyrene) doping concentrations as proxies for device degradation. We fabricated electron-only devices and evaluated the charge transport properties in the photoactive layers with small molecular, oligomeric, and larger-size polymeric acceptors by adopting the space-charge-limited current (SCLC) measurement36,37,38. The chemical structures and energy levels of the selected materials in this work are listed in Fig. 1a–g. Three types of devices were selected as model systems, including small-molecule acceptor-based PM6:Y6, polymer acceptor-based PM6:PY-V-γ, and oligomer acceptor-based PM6:QM1 devices. The electron-only devices with optimized D:A weight ratios for OSCs achieved electron mobility values around 10−4 cm2 V−1 s−1, which were comparable with data reported in the literature39,40,41,42. We also investigated the percolation thresholds of electron transport and impurity tolerance in active layers. Figures 2a–c and S3–S13 show the J–V characteristics of PM6:PY-V-γ, PM6:QM1, and PM6:Y6 films with composition and polystyrene (PS) content variations. The detailed device fabrication process and experimental mechanism have been well documented in the Methods and Supplementary Information (Tables S1 and S2).

Fig. 1: Chemical structures, energy levels, and morphology illustrations.
Fig. 1: Chemical structures, energy levels, and morphology illustrations.The alternative text for this image may have been generated using AI.
Full size image

Chemical structures of (a) polymer acceptors, (b) single-component materials, (c) oligomer acceptor, (d) small molecular acceptors, (e) fullerene acceptor, and f polymer donors. g Energy level diagrams of photovoltaic materials investigated in this study.

Fig. 2: The trend of electron transport properties with different composition content.
Fig. 2: The trend of electron transport properties with different composition content.The alternative text for this image may have been generated using AI.
Full size image

a–c Electron current density × film thickness (J × d) of devices as a function of the applied electric field for PM6:PY-V-γ, PM6:QM1 and PM6:Y6 devices with different PS addition and D:A composition; d–f logarithm of electron mobilities as a function of acceptor weight fraction and various χ values for PM6:PY-V-γ, PM6:QM1 and PM6:Y6 devices, the PS was added at the optimized compositions for the BHJ devices. Source data are provided as a Source Data file.

Firstly, we performed the electron transport measurements to investigate the D:A composition-dependent electron mobilities of the active layers mentioned above, focusing on the formation of the electron percolation process. Figure 2d–f illustrates the logarithm of electron mobilities plotted against the acceptor weight content for three typical BHJ films, involving all-polymer PM6:PY-V-γ, oligomer acceptor-based PM6:QM1, and small-molecule-based PM6:Y6 devices. Most devices with different D:A compositions showed two distinct regions. Initially, there is a sharp increase in electron mobility corresponding to the formation process of the percolating pathway and followed by a gradual rise in electron mobility at higher acceptor weight fractions. The intersection between these two regions can be interpreted as the percolation threshold, indicating the completion of the electron percolation pathways43,44. The percolation threshold is defined as the minimum content of the acceptor material required to form a continuous transport channel in the active layer. Below the percolation threshold, the charge transport network may be fragmented or discontinuous, resulting in reduced charge transport properties. The acceptor weight fractions at the percolation thresholds decrease in the order of small molecular, oligomer, and polymer-based BHJ films. Notably, the all-polymer system containing PM6:PY-V-γ exhibits the largest effective volume fraction χ, indicating polymeric acceptors are effective in transporting electron carriers. Therefore, despite partial acceptor loss from degradation during aging, the polymeric acceptor-based systems retain sufficient components to maintain functionality.

We further introduced PS insulators possessing a linear non-conjugated backbone as additives into the active layers to investigate the charge carrier transport properties and evolution under adverse environments. By adding a low dosage of PS, all electron-only devices exhibited maintainable or even higher electron mobility values than corresponding pristine BHJ films, which can be attributed to the interspace filling effects and the reduction of traps and defects to facilitate the formation of interconnected pathways in the PS-blended films45,46. However, excessive PS content may negatively exceed the impurity content present in the actual device, affect the phase separation of BHJ films, leading to significantly reduced charge carrier mobilities47,48. As the PS content reached 20 wt.%, the range of mobility variation for small-molecule acceptor-based devices changed from 2.1 × 10−4 cm2 V−1 s−1 for pristine devices to 4 × 10−5 cm2 V−1 s−1 for the PM6:Y6 system. In the case of the oligomeric acceptor QM1-based system, a functionalized π-bridge was used as a linker between two non-fullerene A units to enhance the connectivity of the electron transport network. In this configuration, film with 20 wt.% PS shows a favorable electron mobility of 1.1 × 10−4 cm2 V−1 s−1. Polymer acceptor-based devices of PM6:PY-V-γ with better connectivity of electron transport network showed a smoother variation in electron mobility values, implying high tolerance to composition variations and excellent stability in electron transport behavior. Impurities can hinder electron transport via π–π interactions, while the transport process in conjugated long-chain structures remained insensitive due to stable covalent bonds32,49,50. This suggests that polymer acceptor-based systems demonstrate enhanced resistance to impurity-induced degradation while maintaining operational stability.

Degradation of electron transport during light soaking test

The photo stability of electron transport in PM6:PY-V-γ and PM6:Y6 BHJ films was evaluated to elucidate the effect of the connectivity in active layers with similar monomer structures. Figures 3a and S14–S22 illustrate the J–V characteristics of PM6:PY-V-γ and PM6:Y6 electron-only devices with various acceptor compositions. The selection of acceptor weight fractions is primarily based on the percolation process of polymeric materials. The critical point, at around 30 wt.%, represents the acceptor content where the electron percolation network for electron transport is established in polymer acceptors, while the small-molecule acceptors are about to reach the percolation threshold required for establishing such a network. On the other hand, the conditions at 55 wt.% (D:A = 1:1.2) represent the optimal condition for photovoltaic devices and the saturation state for the electron transport network. Figure 3b summarizes the zero-field electron mobilities of the electron-only devices consisting of these acceptors, as a function of the illumination time. Initially, the fresh PM6:PY-V-γ and PM6:Y6 devices with 30 wt.% acceptors achieved electron mobility values of 3.5 × 10−4 and 6.7 × 10−6 cm2 V−1 s−1, respectively. After 70 h of illumination, the electron mobility values of the PM6:PY-V-γ device with 30 wt.% acceptors still maintained 3.5 × 10−5 cm2 V−1 s−1, whereas of the PM6:Y6 case decayed tremendously to only 1.7 × 10−8 cm2 V−1 s−1. As acceptor ratios increase, the electron mobility for both small-molecule and polymer-based devices shows reduced decay, suggesting an improvement in the stability of electron transport behavior due to the increased connectivity. After 70 h, the electron mobility values of PM6:PY-V-γ in optimized ratios decrease from the desirable electron mobility values of 3.9 × 10−4 cm2 V−1 s−1 in pristine devices to residual values of 7.4 × 10−5 cm2 V−1 s−1. With 65 wt.% acceptors, the PM6:PY-V-γ devices maintained residual mobility values of 8.8 × 10−5 cm2 V−1 s−1, exhibiting satisfactory electron mobility stability during the light exposure process. This phenomenon arises from the structural rigidity of the polymer acceptor combined with intramolecular charge transport along the polymer backbone, collectively enabling efficient electron transport through the device. In the case of the small molecules, the PM6:Y6 device with optimized and 65 wt.% acceptors showed inferior transport stability compared to the polymer case and the electron mobility values were only 1.3 × 10−6 cm2 V−1 s−1 and 8.7 × 10−6 cm2 V−1 s−1 after 70 h, due to the vulnerable and unstable percolation pathways, respectively. The detailed electron mobility values are listed in Table S5.

Fig. 3: Degradation of electron transport and photovoltaic performance during light exposure process.
Fig. 3: Degradation of electron transport and photovoltaic performance during light exposure process.The alternative text for this image may have been generated using AI.
Full size image

a Electron current density × film thickness (J × d) of devices as a function of the applied electric field for fresh and aged (standard 1-Sun illumination for 70 h) of PM6:PY-V-γ and PM6:Y6 electron-only devices; b PM6:PY-V-γ and PM6:Y6 electron-only devices in (a) as a function of the standard 1-Sun illumination time; c normalized PCE of PM6:PY-V-γ and PM6:Y6 organic solar cells as a function of the standard 1-Sun illumination time; and d fresh and aged (standard 1-Sun illumination for 70 h) PM6:PY-V-γ and PM6:Y6 electron-only devices with various acceptor weight fractions.

The similar trend was also observed in the organic photovoltaic devices. For both the PM6:PY-V-γ and PM6:Y6 cases, the power conversion efficiencies (PCE) are comparable to the current state-of-the-art OSCs, with values of 16.4% and 16.7% (Fig. S26), respectively. The exposure under illumination and heating resulted in decreased photovoltaic parameters for all devices. The photon stability of both devices was first examined (Fig. 3c), where the PY-V-γ-based device demonstrated a robust PCE value with 82% retained in the final performance after 120 h photo aging, as characterized by the maximum-power point tracking. On the contrary, the small molecular devices expressed inferior photo stability under the same conditions. Meanwhile, the polymer-based devices (PM6:PYF-T-o) and single-component devices (PBDB-T-b-PTY6) maintained around 75% and 64% of their initial PCEs after 500 h thermal aging at 80 °C, respectively, which are higher than small-molecule devices (Fig. S30). The enhanced stability of polymer-based devices is probably owing to the stable charge transport network arising from the rigid backbone conformation.

Figure 3d summarizes zero-field electron mobilities as a function of the acceptor weight fraction for PM6:PY-V-γ and PM6:Y6 devices. We observed that polymer- and small-molecule-based devices with acceptor weight fractions below the percolation threshold exhibited markedly inadequate electrical conductivity and displayed heightened vulnerability to aging-induced degradation. This phenomenon is primarily attributed to the heightened sensitivity of low acceptor weight fractions to adverse effects such as torsion and degradation of the π-conjugated backbone, as well as the increased generation of diverse molecular conformations under light exposure. Therefore, it is crucial to ensure the connectivity of the electron transport network to enhance the electron transport and long-term stability. Notably, aged PM6:PY-V-γ devices can still observe the formation of percolation processes with an increased percolation threshold due to the disruption of the connectivity of the transport network and morphology under illumination. Previous theoretical simulations and optoelectronic experiments have shown that molecules with coplanar and rigid structures can achieve more stable chain conformations51,52,53,54. The monomer units in PY-V-γ are well parallel, together with a more coplanar and rigid conformation. In the case of PM6:Y6 devices, nucleation, growth, and coarsening of crystalline acceptor domains, along with over-purification of mixed domains, resulted in a significant degradation of the electron transport work connectivity among various D:A compositions after 70 h of light exposure, thus impeding the establishment of an efficient transport process. Furthermore, the oligomer-based PM6:QM1 cell exhibited a relatively higher percolation threshold than small molecules with standard π–π stacking transport models. These results indicate the significant effects of connectivity on the electron transport and device stability in well-percolated BHJ films. This suggests a correlation between connectivity and the structure parameter. We investigated the percolation thresholds of blended systems based on PY-V-γ of varying molecular weights (Fig. S31). We observed that reducing molecular weight marginally increases the percolation thresholds in fresh films, showing minimal disparity; however, this difference becomes more pronounced after aging.

Analysis of the electron transport mechanism in blend films

To cover all kinds of OSCs as widely investigated as possible, we expanded the system to four types of films, including small-molecule acceptor-based films of PBDB-T:Y6, PM6:Y6, and PM6:BTP-eC9, polymer acceptor-based films of PM6:PY-IT, PM6:PYF-T-o, and PM6:PY-V-γ, oligomer acceptor-based films of PM6:QM1 and PBDB-T-b-PTY6, and DCPY2 for single-component devices. We evaluated the electron transport properties with respect to impurity tolerance and donor:acceptor weight fractions in the photoactive layers. Figure 4a illustrates the significant reduction in electron mobility values and two distinct partitions of the electron mobility change ratio in devices based on acceptors with diverse structures, when the PS content ranged from 5 to 20 wt.%. Particularly, the range of the mobility variation in the small-molecule acceptor-based devices was the most dramatic, declining from 4.4 × 10−4 cm2 V−1 s−1 in pristine devices to 5.9 × 10−5 cm2 V−1 s−1 in blending films with 20 wt.% PS, and approaching an order of magnitude in the PM6:BTP-eC9 system. Conversely, devices based on polymeric acceptors, oligomeric acceptors, and single-component materials demonstrated excellent connectivity of the electron transport network, resulting in smoother variations in electron mobility. This implied a high tolerance to composition variations and excellent stability in electron transport behavior. However, the hole transport properties of PCDTBT:PC71BM and PBDTTTPD:PC71BM films exhibited remarkable decay when 10 wt.% PS was blended into the BHJ films, as shown in Fig. S25. It indicated that the stability of polymer conformations with excellent connectivity in the electron transport network is not universally applicable and may depend on various factors. Therefore, a more comprehensive quantitative analysis is necessary to fully understand the underlying mechanisms of polymer stability and network connectivity.

Fig. 4: Electron transport mechanism analysis of the blend films.
Fig. 4: Electron transport mechanism analysis of the blend films.The alternative text for this image may have been generated using AI.
Full size image

a Mobility change ratio of devices with different PS content, (b) the electron mobility of devices with different acceptor weight fractions, and c electron percolation thresholds as a function of impurity tolerance for small-molecule acceptor-based polymer acceptor-based films of PBDB-T:Y6, PM6:Y6, PM6:N3, and PM6:BTP-eC9, polymer acceptor-based films of PM6:PY-IT, PM6:PYF-T-o, and PM6:PY-V-γ, oligomer acceptor-based films of PM6:QM1 and the PBDB-T-b-PTY6 and DCPY2 single-component devices. Source data are provided as a Source Data file.

Figure 4b shows the logarithm of electron mobilities plotted against the acceptor weight content for the investigated BHJ devices. The acceptor weight fraction required to achieve electron percolation can be clearly divided into two regions, reflecting differences in the connectivity of the electron transport network. For polymeric networks, an acceptor weight fraction of ~20% is needed for excellent connectivity. For oligomeric systems, this weight fraction increases to ~30%, while small-molecule systems require about 45% weight fraction. Notably, an all-polymer BHJ film of PM6:PY-V-γ attained a sizeable electron mobility (>10−4 cm2 V−1 s−1) even with a small acceptor content of ~20%. Improved carrier mobility was also obtained at 65 wt.%, where electron transport was in a supersaturated state, suggesting that even after electron percolation is complete, more electron conduction pathways are needed in the BHJ to maintain higher carrier mobility. This implies that an all-polymer system can maintain high transport efficiency despite potential degradation of donor/acceptor components during aging.

The electron transport mechanisms differ based on the connectivity of the electron transport network in organic semiconductors. In conjugated polymer materials, local charge transport occurs, coupled strongly to inter- and intramolecular vibrations55. In small molecular amorphous systems, electrons mainly transport via well-formed π–π interactions. Figure 4c clearly illustrates the differentiated effects of π–π stacking transport and the electron transport network model according to the difference in connectivity of the electron transport network in the active layers. Large amounts of insulating polymers usually act as poor conductors for electron carriers, and the mechanisms of inert materials in the active layers have been well documented elsewhere47,48. Herein, we define the impurity tolerance as the ratios of mobility values for films blended with 10 wt.% PS to the mobility values of the neat cases. In small molecular amorphous systems, including PM6:BTP-eC9, PM6:Y6, PM6:N3, and PBDB-T:Y6 films, π–π stacking transport is the dominant transport model. Three all-polymer BHJ films based on PM6 with PY-IT, PY-V-γ, and PYF-T-o as acceptors enjoy desirable electron transport properties, with percolation thresholds/impurity tolerance values of 20%/0.86, 16%/0.42, and 17%/0.51, respectively. The strong intramolecular and intermolecular interactions in these all-polymer systems facilitate electron transport network models. In contrast, the oligomer-based PM6:QM1 device shows a relatively high percolation threshold of 35%, yet achieves an impurity tolerance value of 0.36, surpassing that of the BHJ film using the standard π-π stacking transport model, indicating a critical state between two charge transport mechanisms.

Morphology and crystalline ordering variations in D:A blends

To understand the different electron transport properties from the perspective of crystallinity, we conducted a detailed morphological analysis of organic thin films, including optimized binary films, PS-blended films, and films with various D:A weight ratios by using grazing-incidence wide-angle X-ray scattering (GIWAXS), as depicted in Figs. 5, S23 and S24. In PM6:PY-V-γ and PM6:PYF-T-o blends, the molecular packing models remained nearly constant upon compositional changes, only with slight spacing shift from 3.6 to 3.8 Å. A similar phenomenon was observed in PM6:QM1 and PM6:PY-IT films. On the contrary, in PM6:BTP-eC9 blends, obvious changes happened in both in-plane (111) and out-of-plane (110) diffraction packing directions under either introducing PS or tuning D:A ratios. In addition, the stacking distance d-spacing of (010) peak for optimized BHJ films based on polymer acceptors, as shown in Figs. S28 and S29, and Table S3, decreases from 3.78 Å in fresh films to 3.76 Å in films after thermal aging and even remains unchanged after light aging. On the contrary, the stacking distance d-spacing of (010) peak for optimized BHJ films based on small-molecule acceptors exhibits drastic changes from 3.65 to 3.59 Å after heating and 3.65 to 3.61 Å after lighting, as shown in Figs. S28 and S29, and Table S4. Moreover, the small-molecule acceptor-based BHJ systems also show a dramatic increase in crystalline coherence length. This revealed that the crystalline stacking structure of polymer acceptor blends has more significant stability than that of the NFA blends.

Fig. 5: Morphology of thin films.
Fig. 5: Morphology of thin films.The alternative text for this image may have been generated using AI.
Full size image

a, c Two-dimensional GIWAXS diffraction patterns of corresponding BHJ films. qxy, scattering vector in the in-plane (IP) direction; qz, scattering vector in the out-of-plane (OOP) direction. b, d IP and OOP line cut profiles of the 2D GIWAXS data based on binary and ternary blended films. e 2D GISAXS patterns and f the 1D intensity profiles along the qxy direction of corresponding BHJ films.

We used atomic force microscopy (AFM) to study the surface morphology of real solar cells in Fig. S27. As for the polymer cases, the PY-V-γ-based blend with different acceptor content displays a slightly larger root-mean-square roughness as well as a noticeable fiber network. This facilitates an increase in the connectivity of the electron transport network and the contacting area with electrodes for better charge extraction and electron transport. When 5 wt.% PS as impurities were introduced, the roughness of the small-molecule-based films increased considerably, which implies that the aggregation of molecules was formed on the surface. In addition, AFM height and phase images of these fresh and aged PY-V-γ-based films do not show significant differences in surface morphology, and all films maintained the interpenetrating fiber-like morphology after photo aging. In contrast, the Y6-based films exhibited markedly increased roughness value together with bigger domains after aging, implying that the kinetic aggregation behaviors of Y6 could give rise to the large and directly observed aggregates in the shallow film morphology. The aggregation was even more severe when impurities (5 wt.% PS) were added, resulting in an increase in roughness from 2.51 nm in fresh films to 2.69 nm after aging. The excessive aggregation of the small molecules near the surface leads to large-sized aggregated nanofibers, while polymer-based blends exhibit uniform fibers with stabilized connectivity of the electron transport network, resulting in stable photovoltaic performances.

To better probe the effect on aggregate state size with the aging time, grazing-incidence small-angle X-ray scattering (GISAXS) experiments were performed in Figs. S33–S36 and Table S8. Compared to the freshly prepared PY-V-γ-based blends, the Y6-based BHJ films had a smaller size of clustered acceptor domain 2Rg and intermixed domains, as well as a dense fractal dimension D value, suggesting the higher domain purity and smaller connectivity to the donor domains. Consistent with the morphological changes observed by GIWAXS and AFM, the structural domain sizes of small molecules shifted more pronouncedly after aging compared to the polymer acceptor, with the acceptor domains shrinking from 29.2 nm in the fresh film to 15.6 nm in the photoaged film. Results for the polymer-based systems exhibited smaller changes in crystalline, acceptor domain size, and surface roughness before and after aging, indicating that polymer molecules have smaller diffusion coefficients compared to small molecules. The chain entanglements and connectivity in polymer acceptors and donors suppress molecular diffusion. The observed differences in the stability of electron mobility between different systems can be attributed, at least in part, to variations in their diffusion coefficients and the resulting differences in charge transport pathways.

Contact angle measurements were conducted to assess the surface energy of the materials, aiming to elucidate interactions influencing phase separation and morphology and thereby investigate the different degradation processes observed in the polymer and small-molecule systems (as shown in Fig. S37). The surface energy and Flory−Huggins interaction calculation, as shown in Table S9, displayed a remarkable miscibility between PM6 and PY-V-γ (χPM6, PY-V-γ = 0.044) and a hypo-miscibility in PM6:Y6 (χPM6, Y6 = 0.13) systems. Due to the hypo-miscible nature of the PM6:Y6, it deviates significantly from the efficiency-optimized morphology during OPV aging operation. The larger miscibility in PM6:PY-V-γ systems means a closer approach to the percolation threshold, contributing to a better stability of the device56,57,58. Overall, the high miscibility and low diffusion properties in polymer acceptor blends support our high and stable connectivity of efficient electron transport networks in the operational devices.

The molecular weight of a conjugated polymer is another key factor determining the electrical, morphological, and mechanical properties, as well as its solubility in organic solvents and miscibility with other polymers. The Mn effect of polymer acceptors on the structural, morphological, electrical, and stability was systematically investigated by GIWAXS (Fig. S31), GISAXS, contact angle, and percolation threshold measurement (Fig. S32) for PY-V-γ of varying molecular weights after heating and lighting aging. Mobility measurements in devices with different D:A ratios reveal that the percolation threshold of the fresh PY-V-γ systems does not vary significantly with polymer molecular weight. However, crystalline, molecular packing, acceptor domain size, and miscibility are influenced by the molecular weight. Lower molecular weight results in reduced molecular stacking order, larger domains, and worse miscibility, which is detrimental for device-optimized morphology under aging. We speculate that the large domains observed in lower molecular weight materials might be due to a lack of chain entanglement of this stiff polymer. After 70 h of aging, high molecular weight polymeric materials exhibited more stable changes in crystallization, percolation thresholds, and domain sizes compared to low molecular weight materials. Nonetheless, the performance of low molecular weight materials is superior to that of small-molecule systems due to smaller diffusion coefficients and high miscibility, lending to better connectivity in the charge carrier transport network.

Dynamical simulations for the effect of the connectivity of the electron transport network

To accurately explain the differences in charge transport behavior among devices based on small-molecule, oligomer, and polymer acceptors and the similarities in morphology as the composition changes, we utilized an extended SSH tight-binding model59,60,61,62,63, which highlights the strong electron–lattice interactions in organic semiconductor materials. Detailed modeling information can be found in Fig. S2 and the Supplementary Information. In this work, we focus on the modeling of A-to-A-type J-aggregation formed by coupling the molecular end-groups, which represents a prevalent aggregated morphology in BHJ films, for the three types of systems, including polymer, oligomer, and small-molecule acceptor. Herein, we assume that a negative polaron (i.e., a spatially localized electron charge) is initially generated at the left side of the molecular systems with different numbers of polymerized repeat units (number of repeat units = 4,2,1). During the dynamic simulations of charge transport with the introduction of impurities, an external electric field E along the chain direction of the modeled systems is applied to all systems with the electric field strength E0 = 4 × 105 V/cm, smoothly turned on with time t in the form of a semi-Gaussian form. Increasing the doping amount of PS leads to an increase in the intermolecular aggregation distance d.

We investigated the influence of intermolecular chain distance d [dM1M2 (polymer acceptor) = dM1M2 (oligomer acceptor) = dM2M3 (oligomer acceptor) = dM1M2 (small molecular acceptor) = dM2M3 (small molecular acceptor) = dM3M4 (small molecular acceptor) = d] on the transport rate of negative polarons in the coupling region. Fig. 6a–c, d–f separately describes the evolution of the charge center position xc \(\left[{x}_{{\mbox{c}}}=\frac{{\sum}_{n}a(n-1){q}_{n}}{{\sum}_{n}{q}_{n}}=\frac{{\sum}_{n}a(n-1)({\rho }_{m,n,n}-1)}{{\sum}_{n}({\rho }_{m,n,n}-1)}\right]\) of the negative polaron and the total net charge quantity Qm [\({Q}_{m}(t)={\sum }_{n}{q}_{m.n}(t)={\sum }_{n}{\mbox{e}}({\rho }_{m,n,n}-1)\)] over time under different d values for all systems. The statistics of the transport rate ratios of the polaron centers in the three systems with d increases are based on d = 0.4 nm, as shown in the Fig. 6g. The transmission rate of the polaron center does not follow a strict monotonically decreasing or increasing trend due to the binding effect of the intermolecular coupling region on the polaron. When d < 1 nm, the intermolecular coupling intensity decreases as d increases, making it easier for the polaron center to overcome the binding effect of the coupling region, and increasing the transmission rate. Conversely, for d > 1 nm, the transport rate of polarons in oligomers and small molecules significantly decreases with d increases, while the transport rate in polymer molecules is only slightly reduced, indicating that charge carrier transport in conjugated long-chain structures is less affected. The high crystallinity of small-molecule materials establishes the anisotropic thermally activated hopping mechanism driven by π–π stacking as the primary channel for charge transport. The presence of impurities may occupy the hopping transport sites and obstruct the intermolecular paths, resulting in a significant slowdown in the rate of charge transfer within the small molecules. Notably, better connectivity provides robust electron transport channels, thereby attenuating the effects of impurities in films based on polymer acceptors.

Fig. 6: The dynamical simulations for the effect of the connectivity of the electron transport network in BHJs.
Fig. 6: The dynamical simulations for the effect of the connectivity of the electron transport network in BHJs.The alternative text for this image may have been generated using AI.
Full size image

The evolution of the charge center position xc of the negative polaron in the (a) polymer, (b) oligomer, and c small molecular acceptor-based devices with A-to-A-type J-aggregation structural models under an applied electric field of E0 = 4 × 105 V/cm. The time evolutions of the total net charge quantity Qm (t) (unit: e) in (d) polymer, (e) oligomer, and f small molecular acceptor-based devices, after an applied electric field for E0 = 4 × 105 V/cm. g The statistics of the transport rate ratios of the polaron centers of the three systems as a function of d values. h The evolution of the charge center position xc of the negative polaron in the oligomer acceptor-based devices with different transfer integrals between linking units under an applied electric field of E0 = 4 × 105 V/cm. i The second smallest eigenvalue of the Laplacian matrix corresponding to the graph structures of different molecules is used to represent the algebraic connectivity of molecular structures as a function of weak interaction forces. j Schematic diagram of the remaining electrons after being trapped, with the change in the proportion of acceptors. Red (yellow in small molecular cases) dots represent the donor and acceptor moieties weight ratio of 0.6:0.4, while blue dots represent the donor and acceptor moieties weight ratio of 0.4:0.6.

Conjugated long-chain polymers exhibit robust charge carrier transport properties due to the high intra-chain electronic coupling between the connected molecular units, which establishes continuous transport channels54. To reveal the effect of the linking unit that connects adjacent monomers on the intra-chain transport, we investigated the evolution of the charge center position of the negative polaron with time for different transfer integral (\({t}_{b}\)) cases. The kinetic simulation of the system depicted in Fig. 6h shows that as tb increases, the time required for the negative polaron to overcome the potential barrier at the linking unit decreases, corresponding to an increase in the rate of intra-chain transport. Notably, the negative polaron can still complete efficient transfer through the linking unit even under substantial energy barriers, when the electron transfer integral tb exceeds 0.05 eV. This suggests that effective electronic coupling between multiple molecular units ensures the intra-chain transport of charge carriers, which means that the electron transport network has good connectivity. Moreover, the connectivity of the electron transport network can effectively describe the extent of electronic coupling and communication between adjacent molecular units or sites within an organic semiconductor material. It reflects the intra-chain transport ability of electrons, and the strong interaction between adjacent sites avoids significant scattering or trapping by defects or irregularities. Moreover, both intra-chain transport driven by effective electronic coupling between molecular units and inter-chain transport driven by π–π stacking contribute to the overall charge transport at certain extents in such BHJ films based on polymer acceptors, which constitutes an electron transport network model. It is believed that this electron transport network model exists in highly crystalline semiconducting polymers with small energetic disorders. A strategy to increase the effectiveness of carrier transport in this transport network is to enlarge the transfer integral between the macromolecules by enhancing the planarity and rigidity of the polymer backbone.

To measure the algebraic connectivity of the molecular structure, we selected the second smallest eigenvalue of the Laplace matrix, λ2, as the parameter64,65. We investigated four different molecular graph structures (Fig. S1) with weak interaction forces assigned equal weights, and calculated the λ2 values for different weak interaction forces ranging from 0.1 to 1, as shown in Fig. 6i. The λ2 parameter provides insights into the connectivity and stability of the molecular graph. Comparative analysis reveals that G1 (polymer) exhibits a significantly higher λ2 value than the other three molecular schematics at the same weak interaction force. This implies that G1 possesses superior connectivity and stronger intermolecular interaction forces in comparison to the other molecular structures. Additionally, the minimal cost required to separate G1 into two different subgraphs surpasses that of the other molecular structure maps. These findings contribute to the enhancement of stability.

We analyzed the topological stability of different electronic structures in five categories of electron acceptors, including small molecular acceptors, polymeric acceptors, oligomer acceptors, and single-component materials such as block copolymers and “double-cable” conjugated polymers66,67. The random electron passing test of nanostructured films with various compositions of acceptor moieties was investigated, and the computational details are shown in the Supplementary Information. When large amounts of free electrons were randomly generated and injected into the thin film surface, they would be trapped if there were no ambient acceptor atoms in all allowed directions of movement, as shown in Fig. 6j. The electron transmission efficiency is defined as the percentage of electrons that successfully pass through the system. For systems with excellent connectivity of the electron transport network, including devices based on polymeric acceptors, oligomeric acceptors and single-component materials, the electron transmission efficiency increased from 14% at a low acceptor content (D:A moieties = 0.6:0.4, represented by the red and yellow dots) to 55% at an increased acceptor content (D:A moieties = 0.4:0.6, represented by the blue dots). The small-molecule case exhibited the lowest electron transmission efficiency among different weight ratios due to the undesirable percolation network for electron transport.

Microstructure-stability-charge carrier transport relationships

The schematic diagrams of the evolution of morphology and charge carrier transport properties over an extended aging period are displayed in Fig. 7. Combined with the theoretical calculation results above, we can propose a plausible explanation for the observed morphological and charge transport evolution over time. The different colored clusters represent the donor or acceptor domains. The overlapping region between the function curves represents the transfer integrals between adjacent units in the intra-chain and inter-chain transport model. These transfer integrals provide insights into the strength of electronic interactions and the degree of charge transport between units. The aging process is also accompanied by over-purification of mixed domains, conformational reversal torsion of the π-conjugated backbone, and the formation of impurities or defects in molecular chains or films. These factors can potentially disrupt the connectivity of the electron transport network and the device performance. Long-term aging of small-molecule-based BHJ systems causes significant shrinkage and kinetic aggregation of acceptor domains due to the excellent miscibility with donors, which can destroy the original percolation network for charge transport in the films. Additionally, defects arising from the aging process weaken the electron transfer integral, resulting in the degradation of charge carrier transport properties and device performance. The presence of defects in hopping transport can create more barriers and energy barriers for charge carriers, disrupting the ordered molecular arrangement or introducing additional localized energy levels along the hopping path. These effects trap and recombine electrons or holes, impeding the effective charge carrier transport.

Fig. 7: Microstructure-stability-charge carrier transport relationships.
Fig. 7: Microstructure-stability-charge carrier transport relationships.The alternative text for this image may have been generated using AI.
Full size image

Schematic illustrations of changes in morphology and charge carrier transport properties of the blend films based on small-molecule and polymer acceptors with before and after illumination.

On the other hand, when a polymer acceptor is introduced into the system, the donors and acceptors form a certain degree of phase separation due to the inferior miscibility. After a continuous light/heat aging process of the polymeric acceptor-based films, it was observed that their morphology (including surface morphology and molecular stacking behavior) exhibited enhanced stability compared to films based on small molecules. This desirable stability can be attributed to the excellent connectivity exhibited by the electron transport network within the polymer acceptor materials, with structural rigidity and a smaller diffusion coefficient. For intra-chain transport, when the defects are not located within the conjugated main chain, charge carrier transport can be interrupted by the formation of energy barriers or wells between adjacent transport sites, reducing the transfer integral. Nevertheless, the strong transfer integral in the intra-chain transport model ensures effective intra-chain transport compared to π–π interaction transport. When defects are introduced into the molecular chain and hinder intra-chain transport, alternative intra-chain pathways for the charge carrier transport can still exist in other polymeric molecules, followed by hopping through inter-chain channels. Therefore, the connectivity between donor and acceptor domains is effective for charge carrier transport, as it enables a more robust network resisting local degradation and intrusion of impurities even when some channels are disrupted. The significant differences in the microstructure of thin films, particularly concerning the connectivity of the electron transport network, can be a crucial factor in affecting charge carrier transport properties and photovoltaic performance.

Discussion

This work proposes the impact of connectivity on the stability of OSCs. We systematically evaluated the transport properties in polymer:SMA, polymer:PA, polymer:oligomer, and single-component active layers. Our findings reveal that films exhibiting an electron transport network model demonstrate significantly improved electron percolation properties and a higher tolerance to impurities. Besides the charge carrier transport characterization, this phenomenon can be further explained by random electron passing tests and the algebraic connectivity of molecular structures. The PM6:PY-V-γ and PM6:Y6 BHJ films with similar acceptor carbon skeletons exhibit different photovoltaic performances and transport properties degradation under light exposure, confirming that systems with desirable connectivity enjoy enhanced electron percolation and device stability. Moreover, the planarity and rigidity of the polymer acceptor facilitate the electron transport. Therefore, this work establishes an important principle for achieving stable OSCs with different electron acceptor conformations.

Methods

Materials

Organic photovoltaic materials PM6, PY-IT, BTP-eC9, Y6, PCDTBT, PBDTTTPD, PDIN, and PDINN used in this work were purchased from Solarmer Materials Inc. N3, PYF-T-o, and PY-V-γ were purchased from eFlexPV Limited. PEDOT:PSS (Clevios PVP Al 4083) was purchased from Xi’an Polymer Light Technology Corp. Chloroform, 1,8-Diiodooctane (DIO), and 1-chloronaphthalene were purchased from Sigma Aldrich Inc. The insulating polymer polystyrene (PS) was purchased from Shanghai Titan Scientific Co., Ltd. All materials were used without further purification.

Single-carrier device fabrication and measurements

Electron-only devices. The electron-only devices were fabricated with a conventional configuration of ITO/Al(50 nm)/BHJ/PDIN/Al. The patterned indium tin oxide (ITO)-coated substrates were sequentially cleaned in an ultrasonic bath by using detergent, deionized water, acetone, absolute ethyl alcohol, and isopropyl alcohol for 20 min in each step. These glass substrates were then treated by UV-ozone for 15 min to improve their work function. The 50 nm Al layer deposited onto these cleaned substrates by thermal evaporation is used to block hole carriers. Afterwards, these prepared substrates were transferred into a nitrogen-filled glovebox. The blending solution can be obtained by dissolving organic donor and acceptor materials through an organic solvent such as chloroform (CF). The BHJ active layer films above 200 nm were fabricated by spin-coating the blend solution on top of Al and then annealed. After that, a thin layer of PDIN was spin-coated on BHJ layer to contribute to the electron transport at 5000 rpm for 30 s. Finally, the top Al electrodes of 130 nm were deposited on the top of the PDIN layer by thermal evaporation to finish the preparation of the whole devices, thus yielding the active area of 0.012 cm2 through a shadow mask. The current density-voltage (J–V) characteristic curves of all devices were recorded in a low vacuum environment by employing a computer-controlled Keithley 2612B.

Space-charge-limited current (SCLC) measurements

The dark current characteristics of the electron-only devices under forward bias were extracted to obtain charge carrier mobilities according to the SCLC method by the following equations,

$${J}_{{SCL}}\times d=\frac{9}{8}{\varepsilon }_{0}{\varepsilon }_{r}{\mu }_{0}\theta \exp \left(0.89\beta \sqrt{F}\right){F}^{2}$$
(1)

where JSCL is the space-charged-limited current density, d is the thickness of the tested active layer, ε0 is the permittivity of vacuum, εr is the relative permittivity, θ is defined as the trap factor correlated with free and trapped carrier densities, μ0 is the zero-field carrier mobility, β is the Poole–Frenkel slope and F is the applied electric field.

Organic photovoltaic devices fabrication and measurements

Organic photovoltaic devices with the structure of ITO/PEDOT:PSS/Active layer/PDINN/Ag were fabricated. Patterned ITO substrates were ultrasonically cleaned in acetone, water, acetone, and isopropyl alcohol. Then the ITO substrates were treated with ultraviolet-ozone for 15 min. PEDOT:PSS solution (PEDOT:PSS:deionized water is 1:1) was spin-coated onto the ultraviolet-ozone treated ITO substrates, followed by annealing at 150 °C for 10 min in the ambient conditions. Photoactive blends of PM6:Y6:PYF-T-o with a donor concentration of 7 mg mL−1 were dissolved in a mixed solvent of chloroform:1-chloronaphthalene (99.5:0.5 vol%) and stirred at 40 °C for 4 h. Polymer acceptor-based blend solutions with different weight ratios 1:1.15:0.05, 1:1.1:0.1, and 1:1:0.2 were spin-cast onto the top of the PEDOT:PSS layer at 4000 rpm for 30 s, followed by further thermal annealing at 100 °C for 10 min in the N2 glovebox. For PM6:Y6 and PM6:PY-V-γ devices, donor:acceptor concentration of 7:8.4 mg mL−1 were dissolved in a mixed solvent of chloroform:1-chloronaphthalene (97:3 vol%) and stirred at 40 °C for 4 h and spin-cast onto the top of the PEDOT:PSS layer at 4000 rpm for 30 s, which were further thermally annealed at 100 °C for 10 min in the N2 glovebox. Then the PDINN solution (1.5 mg mL−1 in methanol) was spin-coated onto the active layer at 3000 rpm for 30 s, followed by thermal deposition of Ag (100 nm) top electrode under high vacuum (<2.5 × 10-4 Pa) to fabricate devices with an area of 0.063 cm2. The J–V curves were measured under AM 1.5 G illumination of 100 mW cm−2 by a Keithley 2400 source meter unit. The AM 1.5 G illumination was simulated by a 4-simulator and calibrated by a silicon cell.

Atomic force microscope (AFM) image

AFM measurements were performed on a Bioscope Resolve AFM (Bruker) in a tapping mode under ambient conditions.

Grazing-incidence wide-angle X-ray scattering (GIWAXS) measurement

GIWAXS measurements were performed at beamline 7.3.3 at the Advanced Light Source. The 10-keV X-ray beam was incident at a small grazing angle of 0.11–0.15° to maximize the scattering intensity. The scattered X-rays were detected using a Dectris Pilatus 2 M photon counting detector.

Grazing-incidence small-angle X-ray scattering (GISAXS) measurement

GISAXS measurements were conducted at a Xeuss 3.0 SAXS/WAXS laboratory beamline at Vacuum Interconnected Nanotech Workstation (Nano-X) in China with Kα X-ray of Cu source (operated at 50 kV, 0.6 mA, 1.5419 Å). Samples were prepared on silicon substrates using identical blend solutions as those used in devices.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.