Fully non-fused electron acceptor solar cells with 18% efficiency via a synergistic peripheral substituent strategy

Organic solar cells (OSCs) are a promising photovoltaic technology owing to their light weight, flexibility, and solution-based fabrication process^1,2,3,4,5,6. In recent years, significant advancements in power conversion efficiencies (PCEs) of OSCs have been achieved through the development of novel photovoltaic materials and device fabrication methods^7,8. Particularly, the emergence of non-fullerene electron acceptors with an acceptor–donor–acceptor (A–D–A)-type fused-ring backbone has enabled PCEs exceeding 20% in single-junction OSCs^{9,10,11,12,13,14,15,16}. Nevertheless, these fused-ring electron acceptors (FREAs) suffer from lengthy synthesis routes and low-yield ring-closure reactions, which present great challenges in achieving cost competitiveness and hinder the commercialization of OSCs^8,17,18,19. In this context, non-fused electron acceptors (NEAs) have been proposed, in which the building blocks in conjugated backbone are connected by carbon–carbon (C–C) single bonds^{20,21,22,23,24,25}. This molecular design strategy effectively circumvents complicated and low-yield cyclization reactions, thereby can significantly reduce the cost of photovoltaic materials^26,27. Additionally, the non-fused structures afford diversity for molecular design from various building blocks, which further promotes the development of electron acceptors. Encouragingly, PCE surpassing 19% has recently been achieved for NEAs by delicately regulating active layer morphology²⁸. However, these NEAs still contain complicated fused-ring building blocks, which are far from truly NEAs. When it comes to fully non-fused ring electron acceptors (FNEAs) devoid of any fused-ring structures, there remains a large gap in PCEs between them and state-of-the-art FREAs^{25,29,30,31,32}.

The low PCEs of FNEAs primarily stem from weak acceptor crystallinity and poor active layer morphology, which are caused by C–C single bonds, as shown in Fig. 1a^22,23,33,34. Specifically, the freely rotation of C–C single bonds induces conformational instability of FNEAs, leading to the formation of numerous conformers with varied degrees of torsion^{24,30,35,36,37,38,39}. These instable molecular conformations will weaken the intermolecular π–π interaction, making it challenging to achieve compact and ordered molecular stacking, ultimately resulting in weak acceptor crystallinity. Such unideal stacking further leads to limited delocalization of π-electrons and hinders the electronic coupling between adjacent molecules, thereby restricting light absorption and impeding exciton diffusion and charge transport⁴⁰. Additionally, the chemical structures of FNEAs and polymer donors exhibit remarkable similarity, with all building blocks interconnected by C–C single bonds. Therefore, the blends of polymer donor and FNEA often suffer from excessive miscibility, posing great challenges in achieving a suitable phase separation^41,42,43,44. Moreover, the weak crystallinity of FNEAs also suppresses the formation of fibril network structure due to poor long-range molecular stacking along the specific crystallographic direction³⁴. The lack of high-quality charge transport pathways and the prevalence of carrier recombination^45,46,47,48 further lead to low PCEs, particularly in terms of short-circuit current density (J_sc) and fill factor (FF). Consequently, enhancing acceptor crystallinity and achieving a nano-sized interpenetrating fibril network morphology to realize efficient exciton diffusion and carrier transport are the forefront challenges for FNEA-based OSCs.

a Schematic diagram of weak crystallization and poor fibril network structure of FNEAs. The figure element of FNEAs is adapted from Yang et al.²³ and used under Creative Commons Attribution 4.0 International License (https://creativecommons.org/licenses/by/4.0/). b Molecular design of the fully non-fused electron acceptors in this work. c Summary of the PCEs for FNEA-based OSCs in the literatures and our newly required data. Source data are provided as a Source Data file.

Full size image

Herein, we report four FNEAs (NEH-4F, EEH-4F, NBO-4F, and EBO-4F) through regulating peripheral substituents (Fig. 1b) to address the aforementioned challenges. Specifically, the encapsulated central core endows EEH-4F and EBO-4F with a more planar conformation and improved crystallinity. In contrast, the unencapsulated molecules NEH-4F and NBO-4F suffer from large conformational disorders and poor crystallinity. Furthermore, the extension of alkyl chains on the thiophene π-bridges for EBO-4F fine-tuned the intermolecular stacking and increased surface energy. Upon blending with the semi-crystalline polymer donor PTTz, the enhanced crystallinity of EBO-4F and desired thermodynamic compatibility in PTTz:EBO-4F blend led to a suitable phase separation and pronounced fibrillary network structure. Consequently, a record-breaking PCE of 18.04% associated with a J_sc of 27.9 mA cm^–2 is obtained by the OSC of PTTz:EBO-4F, which are both the champion values for FNEAs to date (Fig. 1c). These results demonstrate the promising prospect of FNEAs in realizing highly efficient yet cost-effective OSCs via rational molecular design and morphology regulation, which will further drive commercial applications of organic photovoltaics in the near future.

Material design and characterization

The chemical structures of the four FNEAs are shown in Fig. 1b, and the corresponding synthetic routes and details are provided in Supplementary Fig. 1. The central cores of FNEAs were constructed by connecting two kinds of different bithiophene units, one with encapsulation structure and the other without, to two flanking thiophene units, respectively. The introduction of the encapsulation structure will effectively inhibit the rotation of carbon–carbon single bond, thereby locking the conformation and enhancing molecular planarity. Simultaneously, the flanking thiophene units were modified by alkyl chains with different lengths to regulate molecular aggregation and donor/acceptor intermolecular interaction. Finally, the target FNEAs were obtained through Knoevenagel condensation between the central core and 2-(5,6-difluoro-3-oxo-2,3-dihydro-1H-inden-1-ylidene)malononitrile (2F-IC). The chemical structures of the intermediates and target FNEAs were confirmed by ¹H and ¹³C nuclear magnetic resonance (NMR), as well as matrix-assisted laser desorption/ionization time-of-flight (MALDI-TOF) mass spectroscopy (MS) (Supplementary Figs. 2−24). NEH-4F, NBO-4F, and EBO-4F exhibited good solubility in common organic solvents, whereas the solubility of EEH-4F in chloroform was observed to be below 5 mg mL^–1, which can be attributed to the short alkyl side chain and excellent backbone planarity of EEH-4F. Thermogravimetric analysis (TGA) presented that the 5% weight loss temperatures (T_d) of NEH-4F and NBO-4F is around 320 °C, but EEH-4F and EBO-4F with encapsulated bithiophene segment show higher thermal stability with T_d exceeding 340 °C (Supplementary Fig. 25a). As shown in Supplementary Fig. 25b, the differential scanning calorimetry (DSC) measurement revealed that NBO-4F exhibits a melting temperature (T_m) of 241 °C and a crystallization temperature (T_c) of 178 °C. Considering the shorter alkyl side chains of NEH-4F and enhanced backbone planarity of EEH-4F and EBO-4F, these acceptors are expected to exhibit stronger crystallinity than NBO-4F. However, neither melting nor crystallization peaks were found for NEH-4F, EEH-4F, and EBO-4F, indicating that the melting temperatures of these molecules may exceed their decomposition temperatures.

Molecular geometries and single crystal analysis

Density functional theory (DFT) calculations were performed on FNEAs to evaluate the molecular geometry (Supplementary Fig. 26). The branched alkyl side chains were replaced by methyl groups to simplify the calculation. We constructed two molecular models to represent the four FNEAs. The atomic coordinates of the optimized computational models are available as Supplementary Data 1. The model FNEA I refers to NEH-4F and NBO-4F with the central bithiophene segment not being encapsulated, while model FNEA II refers to EEH-4F and EBO-4F with the central bithiophene segment being encapsulated. The backbone of FNEA I is twisted with a dihedral angle of 9.36° for the central bithiophene unit, while a more planar molecular backbone is observed for FNEA II with a dihedral angle of only 0.03°. The molecular geometries and intermolecular packing modes of the four FNEAs were further analyzed from single crystals^39,49. All the structure factors are validated by the checkcif test (Supplementary Figs. 27−29 and Supplementary Data 2). As depicted in Fig. 2a−d, the central bithiophene units in NEH-4F, EEH-4F, and EBO-4F exhibit excellent planarity with a dihedral angle of 0°, while NBO-4F displays notable torsion with a dihedral angle of 6.55°. Meanwhile, the dihedral angle between the end group and the thiophene π-bridge in NBO-4F is significantly larger (11.1°) than those of the other three molecules (only 1–2°). These twists impart NBO-4F with a less planar conformation, whereas NEH-4F, EEH-4F, and EBO-4F demonstrate excellent molecular planarity (Fig. 2e−h), thereby facilitating the formation of compact molecular packing. Notably, the phenyl groups connected to the central bithiophene core of all molecules are nearly perpendicular to the conjugated skeleton, which can prevent excessive molecular agglomeration and promote J-aggregate formation. In EEH-4F and EBO-4F, this vertical conformation is stabilized by the encapsulation structure. In NEH-4F and NBO-4F, the rotation of the C–C single bonds between phenyl groups and bithiophene core would alter this vertical conformation, thereby diminishing the aforementioned advantages. Additionally, the distance between the oxygen (O) atom on the 2F-IC ending group and the sulfur (S) atom on the alkylthiophene of the four FNEAs is around 2.7 Å (Fig. 2a−d), which is significantly smaller than the sum of van der Waals radius (r_w = 3.25 Å), indicating the presence of intramolecular S···O = C conformational locks^50,51. Especially, this S···O distance in EBO-4F is the smallest among the four molecules, suggesting the formation of the strongest intramolecular conformational lock.

a–d Top views of the single-crystal structures of NEH-4F⁴⁰, EEH-4F, NBO-4F, and EBO-4F, respectively. e–h Side views of the corresponding single-crystal structures. i–l Crystal network structure and intermolecular packing modes with non-covalent interactions.

Full size image

The crystal network structures of the four FNEAs are shown in Fig. 2i−l and Supplementary Fig. 30. The two-dimensional brick-like packing modes were observed for all FNEAs, which are similar to some conventional A–D–A-type electron acceptors, such as ITIC and its derivatives^{4,52,53,54,55,56}. We extracted the basal stacking patterns of the four acceptors and unexpectedly observed that there was no effective π-π stacking in NEH-4F, which will damage intermolecular charge transport. This abnormal phenomenon can be attributed to multiple non-covalent interactions between adjacent NEH-4F molecules, which facilitate the formation of dimers in the same plane (Supplementary Fig. 31). Moreover, the steric hindrance imposed by the side chains prohibits these dimers from forming π-π stacking between different layers. For the other three FNEAs, there are two intermolecular stacking modes. Packing mode I is characterized by the two pairs of thiophene spacer and the 2F-IC terminal unit, with π–π stacking distances of 3.33, 3.37, and 3.31 Å for EEH-4F, NBO-4F, and EBO-4F, respectively. These values are even smaller than those observed in the state-of-the-art fused ring electron acceptors^57,58. The packing mode II is formed between the neighboring 2F-IC groups with π–π stacking distances of 3.52, 3.64, and 3.65 Å for EEH-4F, NBO-4F, and EBO-4F, respectively. Both two packing modes can provide channels for efficient intermolecular charge transport. Notably, the packing mode I in EBO-4F possesses a larger stacking area than those of the other acceptors, which is beneficial to reducing local structural defects³⁴ and facilitating intermolecular charge transport, leading to higher J_sc and PCE.

Optical and crystalline properties

The optical properties of four FNEAs were investigated by UV-vis-NIR absorption spectra. In dilute chloroform solutions, the absorption spectra of EEH-4F and EBO-4F are indistinguishable due to the elimination of the alkyl chain length effect, as do NEH-4F and NBO-4F (Supplementary Fig. 32). Interestingly, EEH-4F and EBO-4F exhibited a red-shifted and sharper (0-0) absorption peak (748 nm) compared to NEH-4F and NBO-4F (680 nm) in solutions. Notably, the Stokes shifts of EEH-4F and EBO-4F in solution (54 nm) are significantly smaller than those of NEH-4F (101 nm) and NBO-4F (105 nm), as shown in Supplementary Fig. 33. This observation is indicative of a more rigid backbone in EEH-4F and EBO-4F, consistent with the locked and planar conformation induced by the encapsulated bithiophene segment. The same phenomenon has also been found in the as-cast films (Fig. 3a). Compact and ordered J-aggregates have been formed in the as-cast films of EEH-4F and EBO-4F, while NEH-4F and NBO-4F exhibited a loose and disordered stacking in solid state due to their rotatable peripheral phenyl substituents. Furthermore, the trend in Stokes shifts for the four FNEAs in films (Supplementary Fig. 34) aligns with those observed in solutions. Specifically, the Stokes shifts of EEH-4F (144 nm) and EBO-4F (112 nm) are considerably smaller than those of NEH-4F (297 nm) and NBO-4F (309 nm). Upon annealing at 120 °C (Fig. 3b), little changes were observed for EEH-4F and EBO-4F, suggesting that these spontaneously formed J-aggregates are very stable. This feature is conducive to the formation of fibril network morphology in the blend film, thereby promoting exciton diffusion and charge transport. A significant red-shift and obviously increased (0-0) absorption peak was found for NEH-4F film after 120 °C thermal annealing, indicating the transition from disordered agglomeration to ordered J-aggregate. The optical band gap (E_g^opt) of NEH-4F was also reduced from 1.48 to 1.35 eV. Interestingly, NBO-4F exhibited a slight blue-shift with a (0-0) absorption peak at 704 nm after thermal annealing, indicating the discrepant aggregation behavior with respect to the other three molecules. Overall, the results discussed above demonstrate that peripheral substituents can effectively regulate the aggregation states of FNEAs, thereby facilitating a more efficient charge generation and transport process. The electrochemical characteristics of the FNEAs were investigated by cyclic voltammetry (CV) measurements (Supplementary Fig. 35). The electron acceptors with the same molecular skeleton exhibited similar frontier orbital energy levels (Fig. 3c). The highest occupied molecular orbital (HOMO) and lowest unoccupied molecular orbital (LUMO) are –5.62/–3.89 eV for NEH-4F and –5.65/–3.89 eV for NBO-4F, while those of EEH-4F and EBO-4F were determined to be –5.49/–3.91 and –5.54/–3.90 eV, respectively. The slightly down-shifted LUMO levels and more obviously up-shifted HOMO levels of EEH-4F and EBO-4F are consistent with their more planar molecular backbone, which benefit from their encapsulated structures.

a, b UV-vis-NIR absorption spectra of as-cast and 120 °C annealed films of the FNEAs. c Energy levels alignment diagram. d–g Corresponding 2D GIWAXS patterns. h, i The related 1D line-cut profiles. Source data are provided as a Source Data file.

Full size image

The microstructures of acceptor neat films, including molecular orientation, stacking distance (d), and crystalline coherence length (L_c) were investigated by grazing incidence wide-angle X-ray scattering (GIWAXS). The two-dimensional (2D) scattering patterns and the extracted one-dimensional (1D) line-cut profiles in the in-plane (IP) and out-of-plane (OOP) directions are given in Fig. 3d−i and Supplementary Figs. 36−38 respectively. The corresponding lattice parameters are summarized in Supplementary Tables 1−4. As shown in Fig. 3d−g, the prominent (010) diffraction peaks (around 1.85 Å^–1) in OOP direction and (100) diffraction peaks (around 0.44 Å^–1) in the IP direction were found for NEH-4F, EEH-4F, and EBO-4F, indicating a clear face-on molecular orientation. From the 1D profiles (Fig. 3h−i), the π–π stacking distance and corresponding coherence length (L_c) were determined to be 3.45/12.0, 3.37/14.5, and 3.39/22.2 Å for NEH-4F, EEH-4F, and EBO-4F, respectively (Supplementary Table 1). Leaving the outer alkyl chains on thiophene π-bridge unchanged, from NEH-4F to EEH-4F, the reduced π–π stacking distance is consistent with the compact molecular packing caused by the encapsulated molecular skeleton. From EEH-4F to EBO-4F, the steric hindrance enhances along with the extended alkyl chains, resulting in a slight increase of π–π stacking from 3.37 to 3.39 Å. However, the increased L_c enables EBO-4F to form better ordering in long-range, which is beneficial to exciton diffusion and charge transport. The paracrystalline disorder factors (g-factor) of the π-π stacking were calculated to further evaluate the order degree of stacking. The g-factors of 21.4%, 19.2% and 15.6% were obtained for NEH-4F, EEH-4F and EBO-4F, respectively. The decreased g-factor of EBO-4F indicates suppressed paracrystalline disorder, which also promotes exciton diffusion and charge transport. Notably, numerous diffraction spots appear in the 2D scattering patterns of NBO-4F, which is distinctly different from the other three FNEAs. As revealed by the comparative GIWAXS analysis of the as-cast and annealed neat films (Supplementary Figs. 36−38), the initially diffuse diffraction ring evolves into well-defined diffraction spots upon annealing, while a sharp in-plane diffraction signal emerges at 1.57 Å⁻¹, accompanied by an enlarged π-π stacking distance from 3.85 to 4.00 Å (Supplementary Table 3). This feature is consistent well with the spectral blue-shift and sharpened 0-0 vibronic absorption peak of the thermally annealed NBO-4F film (Fig. 3b), which can be attributed to its unstable molecular conformation caused by extended alkyl chains and unencapsulated bithiophene segments.

Mobility and exciton dynamics

Space-charge-limited current (SCLC) experiments were performed to investigate the electron mobilities of these acceptors. As exhibited in Supplementary Fig. 39, the electron mobilities (μ_e) were determined to be 9.70 × 10⁻⁴, 2.38 × 10⁻³, 6.32 × 10⁻⁴, and 2.78 × 10⁻³ cm² V⁻¹ s⁻¹ for NEH-4F, EEH-4F, NBO-4F, and EEH-4F, respectively. The lower μ_e value of NBO-4F highlights the detrimental effect of disorder-oriented weak polycrystalline structure on charge transport. In comparison, the higher μ_es of EEH-4F and EBO-4F are consistent with their optimized aggregate structures and improved long-range ordering, which positively impact charge transport. Furthermore, the exciton lifetime (t) and exciton diffusion length (L_d) of EEH-4F and EBO-4F were measured to investigate the exciton diffusion process via the exciton–exciton annihilation (EEA) method (Supplementary Fig. 40). As listed in Supplementary Table 5, EBO-4F exhibited a longer lifetime of 769 ps compared to EEH-4F (455 ps), providing an extended period for exciton diffusion and leading to a larger L_d value (62.08 versus 57.56 nm). Notably, this L_d is comparable to those of state-of-the-art Y-series acceptors^34,59, ensuring an efficient exciton diffusion process for achieving high J_sc. In addition to influencing the crystallinity, peripheral substituents can also impact the surface energy of acceptors, thereby modifying the thermodynamic miscibility with polymer donors. Consequently, the surface energy of the four FNEAs was measured via contact angle measurements (Supplementary Fig. 41 and Supplementary Table 6). Interestingly, elongating the outer side chain effectively enhances the molecule’s surface energy. This will counterbalance excessive miscibility resulting from FNEA’s similarity to the donor polymer in chemistry, thus contributing to improved phase separation and more refined fibril network structure in the blend films.

Photovoltaic performance and device physics

The photovoltaic performance of the FNEAs was evaluated in a device of ITO/PEDOT:PSS/active layer/PNDIT-F3N-Br:PDINN/Ag. The polymer donors were firstly screened in donor:acceptor binary devices with EBO-4F as the electron acceptor (Supplementary Fig. 42 and Supplementary Table 7). The wide-bandgap polymer PTTz with linear push–pull structure developed by our group previously emerged

as the best choice⁶⁰. The chemical structure, absorption spectrum, and energy levels of PTTz are detailed in Supplementary Fig. 43. The current density–voltage (J–V) characteristics of the optimized OSCs are displayed in Fig. 4a, while the detailed photovoltaic parameters are listed in Table 1. The NBO-4F- and NEH-4F-based devices exhibited a PCE of 12.8% and 14.7%, with an open-circuit voltage (V_oc) of 0.87/0.86 V, J_sc of 19.5/23.4 mA cm^–2, and FF of 75.2%/73.2%, respectively. As shown in Fig. 4b, the PTTz:NBO-4F-based device exhibited the lowest external quantum efficiencies (EQEs) with a response cutoff at approximately 850 nm. These results are consistent with the disordered orientation and weak crystallinity of NBO-4F (Fig. 3f). However, the PTTz:NEH-4F-based device exhibits an extended and enhanced spectral response with EQE exceeding 70%, which is in accordance with the elevated J_sc. The equally high EQE response in both the donor and acceptor regions indicates a comparable charge generation efficiency from the donor and acceptor phases. The devices based on PTTz:EEH-4F and PTTz:EBO-4F demonstrated a lower V_oc (0.82–0.83 V), which agrees well with the down-shifted LUMO energy levels of the corresponding electron acceptors. However, superior J_sc of up to 27.7 and 27.9 mA cm⁻² have been obtained by the OSCs of PTTz:EEH-4F and PTTz:EBO-4F, respectively. The higher J_sc of these two devices can be attributed to the more efficient charge generation, as evidenced by the exceeding 80% EQE (Fig. 4b). The PTTz:EBO-4F-based device also exhibited the highest FF of 77.9%, indicating efficient and balanced charge transport, as well as suppressed recombination. Ultimately, a champion PCE of 18.04% was obtained. Certified PCE reaches 17.48% (Supplementary Fig. 44), marking the highest value for FNEA-based solar cells to date. Notably, this achievement marks the first demonstration of the simultaneous realization of high J_sc (>27 mA cm⁻²) and PCE (>18.0%) in FNEAs (Fig. 4c and Supplementary Table 8). To evaluate the cost-effectiveness, the figure-of-merit (FOM), which is defined as the ratio of PCE and synthetic complexity (SC), was calculated⁶¹. Generally, the higher FOM indicates the more cost-effective of acceptor²⁹. We compared EBO-4F with several representative electron acceptors including Y6⁸, L8-BO⁶², and 2BTh-2F-C₂²⁸ (Supplementary Figs. 45−48), and the detailed parameters are listed in Supplementary Table 9. Notably, EBO-4F achieves a superior FOM of 0.257 compared to Y6 (FOM = 0.229), L8-BO (FOM = 0.229), and 2BTh-2F-C₂ (FOM = 0.211), while maintaining a PCE exceeding 18%. These results highlight EBO-4F as a highly promising low-cost acceptor for future large-scale applications.

a J–V curves. b EQE spectra. c Summary of PCE and J_sc for the reported FNEAs and our data. d TA spectra of the PTTz:EBO-4F blend film under 800 nm excitation at selected delay times. e TA kinetics of the blend films. f Hole and electron mobilities of the blend films. g TPV characterization. h Capacitance–voltage characteristics. i Energy losses of the OSCs.

Full size image

The steady-state photoluminescence (PL) and time-resolved photoluminescence (TRPL) spectra were measured to investigate the charge transfer and exciton dissociation in the OSCs. As shown in Supplementary Fig. 49a, the PL emission of PTTz was almost completely quenched in all blends, which validates efficient electron transfer from donor to acceptor. Moreover, the fluorescence lifetimes (τ) of PTTz in all blends are close to 0.1 ns, suggesting that the electron transfer process is ultrafast (Supplementary Fig. 49b). However, the steady-state PL quenching efficiencies of electron acceptors varied significantly in blends (Supplementary Fig. 50). NEH-4F, EEH-4F, and EBO-4F exhibited a PL quenching efficiency of 86.8%, 87.6%, and 92.3%, respectively, indicating efficient hole transfer in these blends. In contrast, the PL quenching efficiency of NBO-4F was found to be only 76.8% in the blend. Given that NEH-4F and NBO-4F possess similar HOMO levels, the less efficient hole transfer should not be ascribed to the energy level alignment between PTTz and NBO-4F. Instead, this less efficient hole transfer should be attributed to the slow exciton diffusion process as a result of poor acceptor crystallinity and suboptimal phase separation morphology. Femtosecond transient absorption (fs-TA) spectroscopy was further employed to investigate the hole transfer and exciton diffusion dynamics in the blends (Fig. 4d, e and Supplementary Fig. 51). Upon selective excitation of electron acceptors at 800 nm, the ground state bleach (GSB) signals of electron acceptors can be observed in 750–900 nm or 730–850 nm. Subsequently, a new GSB signal attributed to the polymer donor appeared in 550–650 nm within 0.5 ps and reached a maximum within 50 ps, indicating ultrafast hole transfer. By fitting the hole-transfer processes with a biexponential function, two lifetime paramaters τ₁ and τ₂ can be derived (Supplementary Table 10), corresponding to an ultrafast interfacial hole-transfer process and a slow diffusion-related transfer process, respectively. PTTz:EBO-4F exhibited the lowest τ₂ (9.57 ps) and τ_ave (6.07 ps) among the four blends, which suggests a more rapid exciton diffusion from the acceptor phase to the donor/acceptor interface. These results echo the suitable phase separation and excellent acceptor crystallinity in PTTz:EBO-4F blend. The exciton dissociation efficiency (P_diss) of the OSCs was calculated via the plot of photocurrent density (J_ph) versus effective voltage (V_eff) (Supplementary Fig. 52). The device based on EBO-4F showed the highest P_diss value (97.4%) among the four devices (95.9% for EEH-4F, 93.1% for NEH-4F, and 92.8% for NBO-4F), convincing the most efficient exciton dissociation and charge transfer.

The hole mobility (μ_h) and μ_e of the donor:acceptor blends were measured through the SCLC method (Supplementary Fig. 53). Very similar hole mobilities were observed for the four blends (≈3.0 × 10⁻³ cm² V⁻¹ s⁻¹), which benefited from the high crystallinity of PTTz. Only the PTTz:NBO-4F blend exhibited relatively low μ_h, which is consistent with the low EQE response in the donor region. PTTz:NBO-4F also exhibited lower μ_e (1.2 × 10⁻³ cm² V⁻¹ s⁻¹) than the other blends (≥2.4 × 10⁻³ cm²V⁻¹ s⁻¹), agreeing well with the disorder-oriented and weak polycrystalline structure of NBO-4F. Moreover, the shortest and longest charge extraction times were observed for EBO-4F (3.36 μs) and NBO-4F (8.22 μs) (Supplementary Fig. 54 and Supplementary Table 11), respectively. Overall, the blends of PTTz:EEH-4F and PTTz:EBO-4F exhibited relatively high and balanced hole/electron transport among the four blends (Fig. 4f), suggesting the better acceptor crystallinity and more refined fibril network morphology, thereby conducive to the improvement of J_sc and FF in OSCs. The charge recombination behaviors were examined via analyzing the dependence of V_oc and J_sc on light intensity (P_light). Owing to the power-law relationship between J_sc and P_light, i.e., J_sc∝P_light^α, the exponential factor α directly reflects the bimolecular charge recombination degree in OSCs. As shown in Supplementary Fig. 55a, the EBO-4F-based OSC has the highest α value (0.99) among the four devices, suggesting negligible bimolecular recombination in this device. Additionally, the correlation between V_oc and P_light can serve as an indicative measure of trap-assisted recombination. As shown in Supplementary Fig. 55b, the fitted slopes of 1.27, 1.14, 1.45, and 1.04 nkT/q were obtained for the OSCs based on NEH-4F, EEH-4F, NBO-4F, and EBO-4F, respectively. The smaller slope values in EEH-4F- and EBO-4F-based devices imply the suppressed trap-assisted recombination, which contributes to superior J_sc and PCE. The trap-assisted recombination was also investigated by transient photovoltage (TPV)^63,64. As shown in Fig. 4g, the decay of photovoltage exhibited a monoexponential decay behavior. As summarized in Supplementary Table 11, EBO-4F-based device demonstrated the longest carrier lifetime (11.4 µs), while NBO-4F-based device exhibited the shortest carrier lifetime (3.28 µs). These results are consistent with the relationships between V_oc and P_light. The trap density (N_t) in blend films was directly calculated via capacitance–voltage (C–V) measurements (Fig. 4h and Supplementary Fig. 56). The corresponding Mott-Schockley plots and fitted data were summarized in Supplementary Table 12. It can be found that the blends of EEH-4F and EBO-4F offered a lower N_t (3.8–3.9 × 10¹⁵cm^–3) than NEH-4F and NBO-4F (4.9–7.5 × 10¹⁵ cm^–3).

The V_oc variation among the OSCs suggests that peripheral substituents also have an impact on the energy loss (E_loss) of these devices. Therefore, the E_loss of four devices was investigated via highly sensitive external quantum efficiency and electroluminescence measurements (Supplementary Fig. 57). Generally, the E_loss can be determined by the equation:

where E_g is determined by the intersection point of absorption and photoluminescence spectra of the neat acceptor films (Supplementary Fig. 58)^65,66. The overall E_loss was determined to be 0.52, 0.59, 0.61, and 0.57 eV for the OSCs of NEH-4F, EEH-4F, NBO-4F, and EBO-4F, respectively (Supplementary Table 13). The E_loss can be further divided into three parts: the loss in charge generation process (ΔE_CT), radiative recombination loss (ΔE_rad), and non-radiative recombination loss (ΔE_non-rad). As shown in Fig. 4i, the OSCs employed different FNEAs exhibited similar ΔE_CT and ΔE_rad, whereas the ΔE_non-rad differs significantly. According to the equation:

where the ΔE_non-rad can be evaluated by the electroluminescence quantum efficiency (EQE_EL). As shown in Supplementary Fig. 59, the measured EQE_EL values were 1.90 × 10^–4, 6.21 × 10^–5, 1.39 × 10^–5, and 7.70 × 10⁻⁵ for the OSCs based on NEH-4F, EEH-4F, NBO-4F, and EBO-4F, respectively. Thus, the ΔE_non-rad was calculated to be 0.22, 0.25, 0.28, and 0.24 eV, respectively. The relatively large E_loss and ΔE_non-rad in EEH-4F- and EBO-4F-based devices suggest that there is still large room for further PCE improvement.

Aggregation property

The aggregation state and microcrystalline structures of the blend films were investigated by GIWAXS. The two-dimensional patterns and the corresponding one-dimensional line-cut profiles are shown in Fig. 5a−f. The detailed parameters are summarized in Supplementary Tables 14−18. For PTTz:NBO-4F blend, a weak (010) diffraction peak was found at 1.81 Å^–1 in the OOP direction, which can be ascribed to the π–π stacking of PTTz (Supplementary Fig. 60 and Supplementary Tables 14−15). However, the messy diffraction peaks derived from NBO-4F in both OOP direction and IP direction remain in the blend film, which enables the blend film to suffer from disordered orientation and stacking, thereby leading to poor J_sc and PCE in solar cells. In contrast, the NEH-4F-, EEH-4F-, and EBO-4F-based blend films exhibited a clear face-on orientation, along with distinct (010) diffraction peaks in the OOP direction and (100) diffraction peaks in the IP direction. The π–π stacking distance and L_c values were determined to be 3.45/17.35, 3.41/18.97, and 3.38/24.27 Å for NEH-4F-, EEH-4F-, and EBO-4F-based blends, respectively (Fig. 5e, f). Clearly, the reduced π–π stacking distance and increased L_c were obtained in PTTz:EEH-4F and PTTz:EBO-4F, which can be attributed to the increased acceptor crystallinity, thereby resulting in enhanced J_sc and PCE. The g-factors of the (010) diffraction peaks were further calculated to evaluate the degree of stacking ordering. As shown in Fig. 5g and Supplementary Table 16, PTTz:EBO-4F exhibited the smallest g-factor (14.89%), indicating the lowest paracrystalline disorder. This low disorder would facilitate the establishment of a long-range ordered packing in film, thereby promoting the formation of fibril network morphology. In IP direction, each blend exhibits two (100) diffraction peaks, corresponding to PTTz (q ≈ 0.34 Å⁻¹) and the FNEA (q ≈ 0.42 Å⁻¹), respectively. For the diffraction peaks from PTTz, similar lamellar stacking distances were observed for the different blends, but PTTz:EEH-4F and PTTz:EBO-4F exhibited larger L_c and smaller g-factor than that of PTTz:NEH-4F (Supplementary Table 17). The blends of PTTz:EEH-4F and PTTz:EBO-4F also exhibited larger L_c and smaller g-factor for the lamellar stacking in IP direction of the electron acceptors (Fig. 5h and Supplementary Table 18). These results suggest more ordered stacking of both donor and acceptor in the blends with encapsulated electron acceptors, which further promote fibril network formation. Moreover, the largest L_c of acceptor lamellar stacking was observed in the blend of PTTz:EBO-4F, suggesting the positive influence of outer side chains.

a–d 2D GIWAXS patterns. e, f 1D line-cut profiles in OOP direction and IP direction of the PTTz:FNEA blend films. g, h The g-factor and L_c of OOP (010) diffractions and IP (100) diffractions for the electron acceptors in blend films. Source data are provided as a Source Data file.

Full size image

Miscibility analysis

The Flory-Huggins interaction parameters (χ) were calculated to investigate the effect of peripheral substituents on the donor-acceptor interactions⁶⁷. The χ value can be calculated by fitting the depression of T_m (Supplementary Fig. 61) and the volume fraction of the acceptor to a quadratic function using the Nishi-Wang framework and equation:

where T_m⁰ and T_m are the melting points of PTTz as neat film and in blend film. ΔH_f is the melting enthalpy of PTTz. V_m and V_s represent the molar volume of the constitutional unit of PTTz and the FNEA, respectively. φ_s is the volume fraction of the FNEA, and R is the ideal gas constant. Notably, this method only considers the amorphous-amorphous interaction, also referred to as χ_aa. As shown in Fig. 6a, b, the χ_aa values were calculated to be 0.52, 0.76, 1.80, and 1.20 for PTTz:NEH-4F, PTTz:EEH-4F, PTTz:NBO-4F, and PTTz:EBO-4F, respectively. The smaller χ_aa value indicates the better miscibility between donor and acceptor⁶⁷. The smallest χ_aa value of the PTTz:NEH-4F blend suggests the best thermodynamic miscibility between donor and acceptor, thereby minimizing phase separation. Conversely, the largest χ_aa value of the PTTz:NBO-4F blend would lead to significant phase separation. The too large or too small phase separation will hinder exciton dissociation and charge transport, leading to a diminished PCE. Different from PTTz:NEH-4F and PTTz:NBO-4F, PTTz:EEH-4F and PTTz:EBO-4F exhibit moderate χ_aa values, suggesting their potential for forming a more optimal phase separation and a refined fibril network morphology. These results were also verified by the surface contact angle measurements (Supplementary Fig. 62 and Supplementary Table 19). In blends containing semi-paracrystalline polymer PTTz, the crystalline-amorphous interaction parameters (χ_ca) cannot be neglected, which will complement the understanding on the donor/acceptor interactions. Therefore, the χ_ca values were also estimated. Through monitoring the depression in the crystallization temperature of PTTz in blends (Supplementary Fig. 63), the χ_ca values can be obtained from the slope of \((1-{T}_{c}/{{T}_{c}}^{0})\) against \((1-{{\varphi }}_{s})\) according to the equation:

where \({{T}_{c}}^{0}\) and \({T}_{c}\) are the crystallization temperatures of PTTz as neat film and in blends, respectively. The higher χ_ca value suggests the stronger repulsive force between the crystalline regions of PTTz and the amorphous regions of the acceptor, which generally promotes the formation of the purer crystalline ___domain in blends⁶⁷. As plotted in Fig. 6c, d, the χ_ca values of PTTz:NEH-4F, PTTz:EEH-4F, PTTz:NBO-4F, and PTTz:EBO-4F were calculated to be 0.23, 0.28, 0.37, and 0.32, respectively. PTTz:NEH-4F offers the lowest χ_ca among the four blends, indicating excessive amorphous miscibility and poor crystalline ___domain. The highest χ_ca of PTTz:NBO-4F corresponds to its highest χ_aa, suggesting the presence of excessive repulsive force in this blend, which will limit effective inter-crystalline connections. The PTTz:EEH-4F and PTTz:EBO-4F blends exhibited moderate χ_ca values, reflecting suitable donor/acceptor interactions. Considering the improved long-range ordering in PTTz:EBO-4F blend, the slightly higher χ_ca than that of PTTz:EEH-4F blend (0.32 versus 0.28) would benefit of better charge transport between crystalline regions in this blend.

a, b Estimates of χ_aa from the measurements of T_m depression as a function of acceptor volume fraction (φ_s), and χ_aa comparison for the blends. c, d Estimates of χ_ca from the measurements of T_c depression as a function of φ_s, and χ_ca comparison for the blends. e–h TEM images of the optimal blend films. i–l AFM-IR spectra of the optimal blend films. Source data are provided as a Source Data file.

Full size image

Morphology characterization

Transmission electron microscopy (TEM), atomic force microscopy (AFM), and atomic force microscopy-infrared spectroscopy (AFM-IR) were used to further investigate phase separation and fibril network structure of the PTTz:FNEA blends. As shown in Fig. 6e−h, the largest phase separation in PTTz:NBO-4F blend and the smallest phase separation in PTTz:NEH-4F blend can be easily distinguished. And the EBO-4F shows slightly increased phase separation compared to EEH-4F-based blend film. These results are in good agreement with the thermodynamic molecular interactions between PTTz and the FNEA. Furthermore, the AFM-IR measurements were employed to investigate the fibril network structure in blend films (Fig. 6i−l). The wavenumber of 1102 cm⁻¹ was specifically chosen to identify electron acceptors in the blend films (Supplementary Fig. 64). As shown in Fig. 6k, the oversized blocky regions can be observed in PTTz:NBO-4F blend film, which further proves its serious phase separation and indicates poor exciton dissociation. However, the other three blend films all exhibited distinct continuous fibril network morphology. It is worth noting that the fibril in the PTTz:NEH-4F blend film displays blurred boundaries (Fig. 6i), suggesting that the excellent donor/acceptor miscibility impacts ordered stacking of components and the formation of fibril structure. Both PTTz:EEH-4F (Fig. 6j) and PTTz:EBO-4F (Fig. 6l) blend films display a clear fibril network structure, which can be attributed to the enhanced acceptor crystallinity and suitable donor/acceptor miscibility. The fibril network morphology of the blend films can also be observed in the AFM phase images (Supplementary Fig. 65), which allow for the extraction of fibril width. The average fibril widths of NEH-4F, EEH-4F, and EBO-4F-based blend films were determined as 13.8, 14.9, and 15.6 nm, respectively (Supplementary Fig. 66). The larger fibril width of PTTz:EEH-4F and PTTz:EBO-4F blend films substantiates the superior quality of the fibril network structure. This not only facilitates charge transport, but also suppresses charge recombination in the devices, thus affording the higher J_sc and FF. As shown in Supplementary Fig. 67, the PTTz:NBO-4F blend film exhibits a very large root-mean-square roughness (R_q = 3.02 nm), which aligns with its disordered aggregation and excessive phase separation. This unfavorable morphology will adversely impact charge extraction at the interface between the active layer and electrode. The other blend films all possess a smooth surface with R_q below 1.50 nm. Among them, both PTTz:EEH-4F and PTTz:EBO-4F blend films exhibit minimal roughness (≈1.20 nm), facilitating the formation of Ohmic contact for efficient charge extraction. Overall, the analysis demonstrated above clearly elucidates the effects of peripheral substituents on intermolecular interactions and morphology of the blend films based on FNEAs. The suitable phase separation and refined fibril network morphology achieved by PTTz:EBO-4F demonstrate that FNEAs are able to yield optimal active layer morphology and superior device performance comparable to fused-ring electron acceptors.

In conclusion, we report four fully non-fused electron acceptors (NEH-4F, EEH-4F, NBO-4F, and EBO-4F) via regulating peripheral substituents, i.e., encapsulating the central core and extending the outer alkyl side chains. The encapsulation of the central core effectively suppressed the rotation of C–C single bonds, thereby locking the molecular conformation of EEH-4F and EBO-4F. This results in decreased stacking distance, enlarged crystalline coherence length and decreased g-factors, as well as favorable face-on molecular orientation. The extended outer side chains not only improved the solubility of FNEAs, but also endowed EBO-4F with a larger stacking area in solid state, which promotes intermolecular charge transport. Moreover, the extended outer side chains modulated the surface energy of EBO-4F, as well as the thermodynamic miscibility between EBO-4F and the crystalline polymer donor PTTz. As a result, good acceptor crystallinity, suitable phase separation, and refined fibril network morphology were obtained in the PTTz:EBO-4F blend, which thereby led to efficient exciton dissociation and charge transport, as well as suppressed charge recombination. Ultimately, a PCE of up to 18.04% has been achieved, which is the champion value for FNEAs to date. This success demonstrates the promising prospect of the cost-effective FNEAs in realizing high-efficiency solar cells, which will drive the commercialization of OSCs.

Device fabrication

The glass substrates with indium tin oxide (ITO) were cleaned by detergent, sonicated in deionized water, and isopropanol sequentially. After that, the clean substrates were dried in blast oven at 70 °C. The ITO substrates were subjected to oxygen plasma for 5 minutes. An aqueous solution of PEDOT:PSS (Heraeus CLEVIOSTM P VP AI 4083) was spin-coated onto the ITO substrate at 4500 rpm for 30 s, followed by drying at 150 °C for 15 min in air. The substrates were then transferred into a nitrogen-filled glove box. The active layer was fully optimized in terms of donor:acceptor weight ratios, solvents, solvent additives, thermal annealing, and electron transport layers. For the optimal PTTz:NEH-4F devices, PTTz and NEH-4F were dissolved in chloroform (CF) at a total concentration of 9.9 mg mL^–1. For the optimal PTTz:NBO-4F devices, PTTz and NBO-4F were dissolved in CF with 1-chloronaphthalene as the additive (volume fraction of 0.5%) at a total concentration of 9.9 mg mL^–1. For the optimal PTTz:EEH-4F and PTTz:EBO-4F devices, PTTz and the electron acceptors were dissolved in CF with 2,2′-(perfluoro-1,4-phenylene)dithiophene (4FPT) as the solid additive (weight fraction of 1 mg mL^–1) at a total concentration of 9.9 mg mL^–1. The optimal donor:acceptor weight ratio is 1:1.2 for all the blends. The active layers were spin-coated at 4000 rpm to give a thickness of 100 nm. The films were then annealed at 120 °C for 10 min. Afterwards, a 30 nm of PNDIT-F3NBr:PDINN (7:3) was spin-coated on the active layer as electron transport layer. Finally, 100 nm Ag was deposited by thermal evaporation in a vacuum chamber at a pressure of 5 × 10^–6 Torr with a shadow mask.

Measurements and characterization

Nuclear magnetic resonance (NMR)

The ¹H and ¹³C NMR spectra were collected on a Bruker AV-500 MHz spectrometer using tetramethylsilane as a reference in deuterated solvents at room temperature.

MALDI-TOF-MS

MALDI-TOF mass spectrometric measurements were performed on Bruker Daltronik GmbH (autoflex II).

Thermogravimetric analyses (TGA)

TGA was conducted under a dry nitrogen gas flow at a heating rate of 10 °C min⁻¹ on a Netzsch TG 209 F3 apparatus.

Differential scanning calorimetry (DSC)

DSC measurements were performed on a NETZSCH (DSC200F3) apparatus under a nitrogen atmosphere with a heating/cooling rate of 10/20 °C min^–1 for the first cycle and a heating/cooling rate of 20/40 °C min^–1 for the second cycle, respectively.

UV-vis-NIR absorption spectra

UV-visible-NIR spectra were recorded on a Shimadzu UV-3600 spectrophotometer at room temperature. All UV-vis-NIR experiments for solutions were performed in chloroform with the sample concentration of 0.02 mg mL^–1. Films were prepared by spin coating the chloroform solutions onto glass substrates.

Density functional theory (DFT) calculation

The geometry was fully optimized at the B3LYP/6-31 G + (d, p) level of theory by the absence of imaginary frequencies. The long alkyl side-chains were replaced by methyl groups. All calculations were carried out with the Gaussian 16 software.

Cyclic Voltammetry (CV)

CV measurements were performed on a CHI660A electrochemical workstation in a solution of tetrabutylammonium hexafluorophosphate (Bu₄NPF₆, 0.1 M) in acetonitrile at a scan rate of 50 mV s^–1, glassy carbon electrode, platinum wire, and an Ag/AgCl electrode were used as working electrode, counter electrode, and reference electrode, respectively.

Figure-of-merit (FOM) analysis

According to the equation:

where SC represents the synthetic complexity index⁷. There are five parameters: (1) the number of synthetic steps (NSS), (2) the reciprocal yields of the monomers (RY), (3) the number of unit operations required for the isolation/purification of the monomers (NUO), (4) the number of column chromatographic purifications required by the monomers (NCC), (5) the number of hazardous chemicals used for their preparation (NHC). The SC can be calculated according to the following equation:

An empirical coefficient was assigned to each parameter, which accounts for the relative importance. The weight of NSS, RY, NUO, NCC, and NHC is 35, 25, 15, 15, and 10, respectively.

Femtosecond transient absorption spectra (TA)

TA measurements were conducted by a home-built measurement system. This system is driven by a commercial femtosecond (fs) laser with a repetition rate of 1 kHz, pulse duration of ~170 fs and a wavelength of 800 nm. The amplifier (Legend Elite F 1 K HE + II, Coherent, California, USA) is seeded by an oscillator (Mira-HP, Coherent, California, USA) running at 80 MHz. The fundamental laser was split into two beams. One is used to pump a home-built non-collinearly optical parametric amplifier, output of which is used to pump the OPV samples of interest. The pump beam was modulated by a mechanic chopper (MC2000B-EC, Thorlabs, Newton, New Jersey) with a frequency of 500 Hz. The other beam is employed to generate the super-continuum white light, which is used as the probe beam for differential absorption measurements, by focusing onto a sapphire plate. The probing light was guided into a monochromator (Omni-λ200i, Zolix, Beijing, China) and detected by a CCD detector (Pascher Instruments, Lund, Sweden). The time delay between the pump and probe beams is controlled by a mechanical delay line. For TA measurements, all samples were mounted into an optical chamber filled with nitrogen.

Exciton diffusion length measurements

The exciton diffusion lengths (L_ds) were calculated by the equation⁶⁸

where τ is the exciton lifetime and D is the diffusion coefficient. According to the three-dimensional exciton diffusion model, D can be given by \(D=\gamma /8\pi R\), where γ is the bimolecular exciton annihilation rate constant, R is the singlet-exciton annihilation radius that is generally assumed to be 1 nm.

Surface contact angle measurements

The contact angle tests were performed on a Data physics OCA40 Micro surface contact angle analyzer. The surface tension of the materials was characterized and calculated by the contact angles of the water and ethylene glycol via the Owen-Wendt & Kaelble (OW) method, where \(\gamma\)^d and \({\gamma }^{p}\) are the dispersion and polarity components, respectively. The samples were cast on PEDOT:PSS-coated ITO substrates, and the contact angle images were taken when the liquids had been dropped on the sample films for 60 s. The Flory-Huggins interaction parameters (χ) were calculated by the equation:

where \(\gamma\) is the surface free energy of donor and acceptors.

Steady-state photoluminescence (PL)

The PL spectra were recorded on a Shimadzu RF-6000 spectrometer.

Atomic force microscopy (AFM)

AFM images were acquired by using a Bruker Multimode 8 Microscope in tapping-mode. AFM-IR images were acquired by using a Bruker nanoIR3 in tapping-mode.

Transmission electron microscope (TEM)

TEM images were obtained from a JEM-2100F transmission electron microscope operated at 200 kV.

Time-resolved photoluminescence (TRPL)

Fluorescence lifetimes were measured by using the time correlated single photon counting (TCSPC) method (Picoquant HydraHarp 400) and femtosecond laser pulses at 450 nm as the excitation source.

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

Grazing incidence wide-angle X-ray scattering (GIWAXS) experiments were carried out on a Xenocs Xeuss 2.0 system with an Excillum MetalJet-D2 X-ray source operated at 70.0 kV, 2.8570 mA, and a wavelength of 1.341 Å. The grazing-incidence angle was set at 0.20 °. Scattering pattern was collected with a DECTRIS PILATUS3 R 1 M area detector. The raw GIWAXS data were analyzed using Igor 6.37 with a modified NIKA package.

GIWAXS for the comparison of four FNEAs before and after annealing

The Supplementary Figs. 36−38 was acquired on Xeuss 3.0 SAXS/WAXS laboratory beamline at Jianghan University. The X-ray source is a Genix 3D Microfocus Sealed Tube X-Ray Cu-source with integrated Monochromator (30 W). The monochromatic of the light source was 1.5418 Å. The sample was positioned vertically on the goniometer, inclined at 0.2° to the incident ray. In-plane and out-of-plane line-cuts were obtained using Foxtrot-Academic-Edition program.

Paracrystalline disorder factor (g-factor)

The g-factor was estimated using the relationship between the coherence length (L_c, derived from Scherrer analysis) and the broadening of diffraction peaks in weakly disordered systems^69,70:

Considering the interplanar distance d, the g-factor can be estimated and defined as:

Δq is the full width at half the maximum of a diffraction peak, defined as:

Combining the two equations above, the g-factor is obtained as:

Device characterization

External quantum efficiencies (EQEs)

The EQE spectra were acquired on a commercial EQE measurement system (Enlitech Co., Ltd. QE-R3011).

Light-intensity dependence measurements

The light-intensity dependence measurements were carried out with illumination between 10 and 100 mW cm^–2, which was calibrated by a standard single-crystal silicon solar cell (Enlitech). The current density and voltage were recorded with a Keithley 2400 source meter.

Photovoltaic performance measurements

The photovoltaic performances were measured under AM 1.5 G irradiation (100 mW cm^–2) derived from a class solar simulator (Enlitech, Taiwan), which was calibrated by a China General Certification Center-certified reference single-crystal silicon cell (Enlitech). The current density–voltage (J – V) curves were recorded with a Keithley 2400 source meter. The device area is 0.0516 cm^–2, and the test was performed with a mask aperture which defined an effective area of 0.04 cm^–2.

Fabrication and characterization of single-carrier devices

The charge mobilities were measured in single-carrier devices with a structure of ITO/ZnO/active layer/PNDIT-F3NBr/Ag for electron-only devices and a structure of ITO/PEDOT:PSS/active layer/MoO₃/Ag for hole-only devices. The dark current densities were measured by applying a voltage between 0 and 4 V with a Keithley 2400 source meter.

Charge carrier mobility estimation

The charge carrier mobility was estimated by fitting the data acquired from single-carrier devices to a space-charge-limit-current (SCLC) model. The mobility was determined by fitting the dark current according to the Mott-Gurney law that consider a Poole-Frenkel-type dependence of mobility on the electric field, given by the following equation:

where J is the dark current density, ε₀ is the permittivity of free space, ε_r is the dielectric constant of the polymer which is assumed to be 3 for organic semiconductors, μ₀ is the zero-field mobility, γ is a parameter that describes the strength of the field-dependence effect, V is voltage drop across the device, and d is the thickness of the active layer. The hole and electron mobilities are extracted with the fit parameters at an electric field (E) of 1.0 × 10⁵V cm^–1 by the Murgatroyd equation:

Reporting summary

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