20.46% efficient organic solar cells with concurrent voltage enhancement and thermal stability enabled by crystallization-kinetics-controlled morphology

Abstract

Organic solar cells are often limited by morphological instability and suboptimal phase separation, largely stemming from the rapid crystallization kinetics of state-of-the-art non-fullerene acceptors, which lead to excessive aggregation and metastable blends. Herein, we design an asymmetric acceptor, BTP-FClO, featuring slowed nucleation dynamics. When incorporated as a third component into PM6:L8-BO blend, BTP-FClO functions as a crystallization moderator, significantly delaying nucleation and phase separation. The refined morphology of the ternary blend is characterized by an extended nucleation time (199 ms) and a prolonged carrier lifetime, resulting in a low energy disorder (13.8 meV) and a lower trap density (2.69 × 10¹⁶ cm⁻³). Thus, the ternary device overcomes the voltage–current trade-off and provides a power conversion efficiency of 20.46% by simultaneously increasing the open-circuit voltage (0.916 V) and short-circuit current density (28.02 mA cm⁻²). Notably, an efficiency of 18.28% is retained even at an active-layer thickness of 446 nm, underscoring excellent thickness tolerance. Moreover, the ternary blend exhibits exceptional thermal stability, retaining 80% of its initial efficiency after annealing for 448 h at 60 °C, attributed to its robust morphology and high glass transition temperature (T_g = 120 °C). This work demonstrates that molecular design targeting crystallization kinetics, alongside energetics, offers a practical pathway toward high-performance organic photovoltaics.

---

Broader context

Organic solar cells face a fundamental challenge in reconciling high open-circuit voltage with efficient charge generation. A sufficient energetic offset at the donor–acceptor interface is necessary to drive exciton dissociation, yet this inevitably compromises the achievable voltage. Conversely, the excessively rapid crystallization kinetics of state-of-the-art non-fullerene acceptors often lead to unfavorable phase separation and metastable morphologies, further limiting device performance and stability. The ternary strategy offers a versatile platform for addressing these issues, where a selected third component can modulate energy alignment, broaden absorption, and refine blend morphology. However, developing guest materials that concurrently regulate the electronic structure and film-forming kinetics remains difficult. To address this, we designed BTP-FClO, an end-group-asymmetric non-fullerene acceptor engineered to weaken intramolecular charge transfer and suppress rapid aggregation synergistically. This dual design yields an elevated lowest unoccupied molecular orbital level and slowed nucleation dynamics. When incorporated into the PM6:L8-BO host, ternary devices achieve simultaneous enhancements in both open-circuit voltage (0.916 V) and short-circuit current density (28.02 mA cm⁻²), overcoming the voltage–current trade-off to achieve an efficiency of 20.46% that outperforms the binary counterpart. Notably, these ternary devices exhibit exceptional thermal stability, retaining 80% of their initial performance after 448 hours at 60 °C.

Introduction

Solution-processed organic solar cells (OSCs) have attracted considerable research interest owing to their low cost, mechanical flexibility, and lightweight nature.^1–6 Recently, immensely advanced active-layer materials, including both polymer donors and non-fullerene acceptors (NFAs), especially Y-series acceptors, have achieved power conversion efficiencies (PCEs) beyond 20%,^7–11 showing promising industrialization prospects for building-integrated photovoltaics and portable electronic devices. Yet, as excitonic solar cells, OSCs commonly adopt a bulk heterojunction structure to overcome exciton dissociation and transport limitations arising from the low dielectric permittivity of organic materials.^12–16 For example, PM6:L8-BO and PM6:BTP-eC9, as two successful active layer combinations, guarantee effective exciton dissociation owing to the sufficient energy offset between donor and acceptor materials, leading to high short-circuit current density (J_SC > 27 mA cm⁻²).^17–21 However, state-of-the-art NFAs, such as L8-BO and BTP-eC9, despite their high efficiencies, inherently exhibit excessively rapid crystallization kinetics, leading to uncontrolled molecular aggregation, poorly defined phase separation, and ultimately metastable morphologies with suboptimal donor/acceptor networks. Consequently, such morphological instability limits further improvement in PCE and undermines device operational stability. Thus, strategically modulating nucleation kinetics to guide the formation of a thermodynamically favorable, robust nanoscale morphology remains a key challenge for realizing high-performance OSCs.^22–25

The ternary strategy, which incorporates a guest component into binary donor–acceptor blends, has emerged as a practical approach to resolve the dilemma mentioned above.^26–30 For the exploration of the third component, a larger bandgap NFA was suitable as the guest component to achieve cascade energy level alignment, a panchromatic absorption spectrum, and a more favorable blend nanomorphology.^31–34 This approach can potentially maximize charge generation in ternary OSCs, thereby improving device performance. Strategic incorporation of a guest NFA has been used to tailor the energetic alignment at the donor/acceptor interface; a guest component with a higher highest occupied molecular orbital (HOMO) level enables balanced charge generation and transport.^35,36 In a related approach, an oligomeric acceptor was introduced as a guest to modulate intermolecular interactions and suppress excessive aggregation of the host acceptor. Similarly, a highly luminescent guest acceptor has been integrated into a binary system to reduce non-radiative energy loss.³⁰ However, these approaches primarily address electronic or photophysical properties after film formation, with little focus on how guest components influence the kinetics of active-layer morphology development during processing. Notably, an exception is the use of an asymmetric brominated acceptor (T10) as the guest, which prolongs the nucleation time, slows overly rapid crystallization, and promotes the formation of an optimized fibrillar network.³⁷ Combining these concerns, we propose that the development of dual-role guest components that simultaneously modulate optical properties, energy alignment, and film-forming kinetics should be a central focus in the design of high-performance ternary OSCs.

In this work, we design and introduce BTP-FClO, an end-group-asymmetric NFA with an elevated lowest unoccupied molecular orbital (LUMO) level and prolonged nucleation kinetics, as a guest component in the PM6:L8-BO blend. The synergistic effects of reducing halogenation and introducing a methoxy group (an electron-donating unit) at the end groups lead to a weak intramolecular charge-transfer (ICT) effect and intermolecular interaction, resulting in an enlarged bandgap, an up-shifted LUMO, the glass transition temperature (T_g), and a prolonged nucleation time twice than that of L8-BO. Besides, the structural similarity between the guest component and the host acceptor ensures good compatibility between the two acceptors, which is critical for achieving a well-mixed yet phase-separated morphology and a high fill factor (FF) in ternary OSCs, and as a result, incorporating BTP-FClO as a third component prolonged nucleation kinetics, enabling the formation of a well-defined nanofibrillar network with optimal domain size. Ternary OSCs based on PM6:L8-BO:BTP-FClO achieved a PCE of 20.46% (certified: 20.12%) and exhibited enhanced operational stability, attributable to concurrent improvements in V_OC, J_SC, FF, and T_g. Notably, ternary OSCs retain high efficiency (>18%) even at active-layer thickness approaching 450 nm. Comprehensive photophysical and morphological analyses revealed that BTP-FClO incorporation extended the carrier lifetime to 6.93 µs and reduced the trap density to 2.69 × 10¹⁶ cm⁻³, resulting in a lower charge recombination rate and a higher FF. This work demonstrates that strategic modulation of nucleation kinetics provides a powerful approach to designing guest components that optimize morphology and enhance performance in ternary organic solar cells.

Results and discussion

The chemical structures of PM6, L8-BO, and BTP-FClO are presented in Fig. 1a. Detailed synthesis procedures and characterization, including nuclear magnetic resonance and mass spectra of BTP-FClO, are provided in Fig. S1–S4. Quantum chemical calculations at the B3LYP/6-31G(d,p) level of density functional theory (DFT) were applied to investigate the differences in energy levels and dipole moment between BTP-FClO and L8-BO. As shown in Fig. S5, since BTP-FClO is composed of multiple isomers, we calculated two configurations for BTP-FClO. The calculated energy levels (LUMO/HOMO) of BTP-FClO are −3.59/−5.66 and −3.57/−5.63 eV, and that of L8-BO is−3.67/−5.73 eV. The up-shifted energy levels and enlarged bandgaps of BTP-FClO are attributed to the weak ICT effect between the backbone and end groups, arising from the synergistic effects of fluorination reduction and the incorporation of the methoxy group, which are beneficial for achieving higher V_OC in the devices.³⁸ Additionally, the asymmetric BTP-FClO shows larger dipole moments (6.34/7.37 Debye) than that of L8-BO (1.35 Debye) (Fig. S6), which has been reported as a key factor in enhancing the dielectric constant, elevating the LUMO level, and reducing non-radiative recombination loss.^39,40 Besides, we conducted all-atom molecular dynamics (AA-MD) simulations (Fig. S7) on these acceptors to explore the assembly deduction. The BTP-FClO bi-molecule delivers a non-covalent interaction energy (E_i) of −57.83 kcal mol⁻¹, slightly lower than that of −59.64 kcal mol⁻¹ for the L8-BO bi-molecule, confirming that the 2F–F and the steric effect of the methoxy substitution on the end group reduce the self-aggregation behavior of molecules.

Fig. 1 (a) Chemical structures of PM6, BTP-FClO, and L8-BO. (b) Normalized absorbance spectra of solutions (dashed lines) and thin films (solid lines). (c) The energy levels of the materials. (d) The out-of-plane (OOP) of three neat films. (e) 2D time-resolved in situ absorption spectra of different acceptors during the solution-to-film transformation process.

UV-vis absorbance spectra were recorded to investigate the absorption properties of these materials in both solution and thin-film states (Fig. 1b). In solution, the maximum absorption peaks (λ_max,sol) of BTP-FClO and L8-BO are located at 725 and 728 nm, respectively, indicating a weaker ICT effect in BTP-FClO, consistent with the trend of DFT results (Table 1). In thin films, all acceptors exhibit bathochromic absorption relative to their solutions. The asymmetric BTP-FClO possesses a smaller value of Δλ_max (Δλ_max = λ_max,film − λ_max,sol, 62 nm for BTP-FClO) than that of the symmetric L8-BO (74 nm), which is consistent with the above E_i results. Besides, the optical bandgaps (E_g) of BTP-FClO and L8-BO in thin-film states are 1.51 and 1.46 eV, respectively, determined from their absorption onsets of these two acceptors. Thus, by combining ultraviolet photoelectron spectroscopy (UPS) characterization (Fig. S8 and S9, SI) with the determined HOMO levels, the energy landscape of the three materials is depicted in Fig. 1c and Table S1. The LUMO and HOMO levels shift from (−4.33 eV, −5.79 eV) of L8-BO to (−4.22 eV, −5.73 eV) of BTP-FClO. Both materials are supposed to match well with PM6, whose energy level is located at ca. (−3.54 eV, −5.43 eV). In addition to neat films, the UV-vis absorbance spectra of the blend acceptors are shown in Fig. S10. The blend film follows a trend similar to that of the small molecules; however, only the blend with a 1 : 0.2 ratio exhibits a minimal blueshift while simultaneously enhancing absorbance at the 710 nm shoulder. Therefore, a 1 : 0.2 acceptor ratio (L8-BO:BTP-FClO) was selected, potentially improving both J_SC and V_OC. The ternary blend exhibits enhanced absorbance in the 600–800 nm region compared to the PM6:L8-BO binary blend in Fig. S11. This enhanced absorbance can be attributed to the complementary absorbance of BTP-FCIO and L8-BO.

Table 1 Optical and electrochemical properties of L8-BO and BTP-FClO

Material	λ max,sol (nm)	λ max,film (nm)	Δλmax (nm)	λ onset,film (nm)	E g (eV)	HOMO/LUMO^b (eV)	Dipole moment (Debye)
a Calculated from the absorption onset of the films. b Estimated from UPS results and optical bandgaps.
L8-BO	728	802	74	849	1.46	−5.79/−4.33	1.35
BTP-FClO	725	787	62	821	1.51	−5.73/−4.22	6.34/7.37

---

The above-mentioned properties of these materials are intimately linked to the molecular packing in the thin films. Therefore, we employed two-dimensional grazing-incidence wide-angle X-ray scattering (GIWAXS) to investigate the differences in molecular stacking in the neat films of PM6, L8-BO, and BTP-FClO. The 2D GIWAXS patterns and corresponding 1D line profiles are presented in Fig. 1d and S12 and S13, with quantitative parameters summarized in Table S2. All the acceptors exhibited a perfect face-on orientation. Moreover, due to the steric effect imposed by the methoxy groups, BTP-FClO demonstrated a similar π–π stacking distance (3.36 Å) and a smaller crystallite coherence length (CCL = 22.99 Å) compared to that of L8-BO. The time-resolved two-dimensional absorption mapping of spin-coated films for different acceptors in chloroform (CF) solution is shown in Fig. 1e and S14. As shown in Fig. S15, the kinetic curves indicate that incorporating BTP-FClO prolongs this process, extending the time window for ordered molecular assembly. With the addition of BTP-FClO, the blended acceptors exhibit enhanced absorption intensity at wavelengths beyond 750 nm. This increase in absorption indicates that BTP-FClO effectively enhances the system's photon capture efficiency in the visible region, thereby promoting more efficient charge generation. The neat L8-BO film exhibited a characteristic timescale for a rapid crystallization of 40 ms. In contrast, the low aggregation tendency of BTP-FClO resulted in slower crystallization kinetics (80 ms; see Fig. S16). Notably, upon incorporating BTP-FClO into the L8-BO host at ratios of 1 : 0.2, 0.8 : 0.4, and 0.4 : 0.8, the crystallization timescale progressively increases to 66 ms, 71 ms, and 76 ms, respectively. This systematic slowing effect confirms that BTP-FClO acts as a kinetic modulator, suppressing excessive phase separation and facilitating the formation of finer domains. The ratio 1 : 0.2 emerges as the optimal formulation, achieving a balance between the necessary crystallization delay and the maintenance of charge-transport pathways. However, excessive BTP-FClO content (>0.2) leads to overly sluggish kinetics, potentially resulting in molecular disordering and reduced charge carrier mobility, thereby compromising device performance. These results demonstrate that BTP-FClO can act as an effective crystallization kinetics modulator. By delaying the crystallization rate of the acceptor phase, it achieves precise regulation of the phase-separated morphology in the blend film.

To investigate the influence of BTP-FClO on the crystallization kinetics of the active layer, we employed in situ UV-vis absorption spectroscopy to monitor the film-formation dynamics during spin-coating (Fig. 2a). Generally, the film-forming process comprises three main stages: the solution state, the solution-to-solid transition state, and the solid thin-film formation state.^41–45 In the first stage, corresponding to the initial solution state, the absorption peaks of BTP-FClO remain unchanged, with the absorption signals primarily reflecting the molecular dispersion characteristics of the solvent. The second stage is accompanied by rapid solvent evaporation, during which molecules begin to aggregate and form initial nucleus structures, leading to distinct red or blue shifts in the absorption features. The third stage corresponds to the formation of the final film morphology.

Fig. 2 (a) 2D time-resolved in situ absorption spectra of different active layers during the solution-to-film transformation process. (b) Dynamic response spectra of the material following photoexcitation at various delay times. (c) Temporal evolution of peak positions for different active layers during the spin-coating process.

The time-dependent absorption spectra in Fig. 2b, together with the quantitative evolution of characteristic peak intensities in Fig. 2c, consistently confirm the regulatory effect of BTP-FClO on crystallization kinetics. Quantitative analysis reveals that while the binary blends PM6:BTP-FClO and PM6:L8-BO exhibit comparable nucleation onset times, incorporating BTP-FClO into PM6:L8-BO to form a ternary system significantly delays the acceptor-phase nucleation time from 132 ms to 199 ms. This change mirrors the intrinsically slower crystallization of neat BTP-FClO but is further amplified in the blend through specific intermolecular interactions. By raising the energy barrier for the ordered molecular packing of L8-BO, these interactions suppress its nucleation threshold, thereby extending the processing window for molecular diffusion and self-organization in the liquid state.

The optimal weight ratio between L8-BO and BTP-FClO was validated by contact angle analysis, as depicted in Fig. S17, S18 and Table S3 (SI). The surface free energies (γ_S) of PM6, L8-BO, and BTP-FClO were 19.81, 24.20, and 25.29 mJ m⁻², respectively. BTP-FClO exhibits higher hydrophilicity, smaller contact angles, and higher surface free energy, while the contact angles and surface free energy of L8-BO fell between those of the other two. When the ratio of L8-BO to BTP-FClO was 1 : 0.2, the Flory–Huggins interaction parameter (χ) for PM6:(L8-BO:BTP-FClO) was 0.0245 K, indicating that the addition of BTP-FClO adjusts the miscibility between the donor and the host acceptor, thereby improving the multi-scale morphology of the blended film.

Subsequently, the wetting coefficient was estimated for the PM6:L8-BO:BTP-FClO ternary blend system to predict the location of BTP-FClO in the blended film. The wetting coefficient of material C (ω_C) in a blend of material A and material B is calculated using Young's equation:^46–49

where γ_X–Y is the interface surface energy between X and Y. If ω_C < −1, C locates in domain B; if ω_C > 1, C locates in domain A; if −1 < ω_C < 1, C locates in the interface between A and B. The calculated wetting coefficient (ω_BTP-FCIO = 1.46) indicates that BTP-FCIO preferentially resides within the L8-BO acceptor phase. This localization allows BTP-FCIO to directly modulate the crystallization kinetics of the host acceptor, effectively prolonging the nucleation time and promoting a more favorable nanomorphology.

To further investigate the nanoscale phase-separated structures, we conducted nanoscale characterization of the surface morphology of the blend films using atomic force microscopy (AFM) and transmission electron microscopy (TEM), with the results shown in Fig. 3a and b. The root-mean-square roughness (R_q) values of PM6:BTP-FClO, PM6:L8-BO, and PM6:L8-BO:BTP-FClO are 0.926 nm, 0.997 nm, and 0.959 nm, respectively, indicating that the incorporation of a small amount of BTP-FClO (at a ratio of 1 : 0.2) does not disrupt the initially favorable morphology of PM6:L8-BO. Meanwhile, the TEM images clearly reveal donor–acceptor networks with distinct structural lengths. The BTP-FClO-based binary films exhibit relatively large phase-separated D–A networks and weaker molecular aggregation. The observed morphology aligns with the reduced self-aggregation tendency of BTP-FClO, which is attributed to the steric hindrance of its methoxy groups. The incorporation of a small amount of BTP-FClO helps suppress excessive crystallization of L8-BO, leading to a more favorable, balanced nanoscale phase separation. This optimized morphology facilitates efficient charge separation and transport, thereby enhancing the device performance.

Fig. 3 (a) Atomic force microscopy (AFM) images, (b) transmission electron microscopy (TEM) images, and (c) the 2D GIWAXS patterns of the three blend films and the corresponding (d) IP and (e) OOP line-cuts.

To explore the underlying reasons for performance improvement from the perspectives of molecular packing and crystallinity, we analyzed the blend films using grazing-incidence wide-angle X-ray scattering (GIWAXS), with results shown in Fig. 3c–e and Tables S4 and S5. The BTP-FClO-based binary films exhibit a diffraction peak of BTP-FClO (q_xy = 1.86 Å⁻¹), corresponding to a d-spacing of 3.38 Å and a CCL of 22.18 Å. The L8-BO-based binary films exhibit a d-spacing of 3.33 Å and a CCL of 26.43 Å. Furthermore, the (010) π–π stacking diffraction peak of PM6:L8-BO:BTP-FClO is located at q_xy ≈ 1.92 Å⁻¹, corresponding to a d-spacing of 3.27 Å, which is smaller than those of the two binary systems. This suggests closer π–π stacking and stronger intermolecular interactions in the ternary system, enhancing lateral charge transport and improving carrier mobility. GIWAXS analysis reveals, at the molecular level, that the introduction of BTP-FClO optimizes the molecular packing order of the acceptor phase, induces tighter π–π stacking and longer-range crystalline domains, and constructs a more favorable mixed molecular orientation. These nanoscale structural advantages collectively establish a robust foundation for highly efficient and balanced charge transport in the device, serving as the key structural factor that ultimately enables ultrahigh efficiency.

The favorable morphology directly suggests improved charge transport. To verify this, we systematically evaluated carrier mobility using photoinduced charge-carrier extraction by linearly increasing the voltage (photo-CELIV) and the space-charge-limited current (SCLC) method. The photo-CELIV technique directly measures the total carrier mobility in the photoactive layer, more accurately revealing the device's operational mechanisms during operation. As indicated in Fig. S19, the peak current (Δj) emergence time (t_max) of the PM6:L8-BO:BTP-FClO ternary device occurs much earlier than that of the two binary devices. In photo-CELIV measurements, a shorter t_max indicates higher carrier mobility and faster transport speed.⁵⁰ Furthermore, the total carrier mobilities of PM6:BTP-FClO, PM6:L8-BO, and PM6:L8-BO:BTP-FClO are 2.11 × 10⁻⁴, 3.14 × 10⁻⁴, and 4.86 × 10⁻⁴ cm² V⁻¹ s⁻¹, respectively. Conversely, the SCLC model enables the independent determination of hole and electron mobilities.

As presented in Fig. S20 and Table S6, the μ_e/μ_h values of PM6:BTP-FClO, PM6:L8-BO, and PM6:L8-BO:BTP-FClO are 5.87 × 10⁻⁴/4.74 × 10⁻⁴, 8.26 × 10⁻⁴/7.51 × 10⁻⁴, and 1.20 × 10⁻³/1.13 × 10⁻³ cm² V⁻¹ s⁻¹, respectively, with the ternary device demonstrating more balanced carrier transport with a ratio of 1.06. The consistent trends from both measurement techniques indicate that the dipole moment modulation induced by BTP-FClO facilitates the formation of a stable network structure, thereby promoting efficient charge transport.

To investigate the effect of BTP-FClO as a guest component on device performance, we fabricated devices with a conventional structure of ITO/PEDOT:PSS/active layer/PDIN/Ag, using PM6 as the donor. The J–V curves are shown in Fig. 4a and b and Fig. S21, and key photovoltaic parameters are summarized in Table 2 and Table S7. Benefiting from its high-lying LUMO level, the PM6:BTP-FClO binary device achieved an exceptionally high V_OC of 0.990 V, though its relatively narrow absorption led to a moderate PCE of 15.13%. Notably, the ternary device that resulted from adding BTP-FClO to the PM6:L8-BO host achieved an improved V_OC of 0.916 V in comparison to the PM6:L8-BO binary device (0.890 V), as well as improved J_SC (28.02 mA cm⁻²) and FF (79.72%). The V_OC enhancement stems from the elevated LUMO of BTP-FClO, while its complementary absorption contributed to the high J_SC. As a result, the PM6:L8-BO:BTP-FClO ternary device delivered a champion PCE of 20.46%, ranking among the highest efficiencies reported for Y-series acceptor-based ternary OSCs (Tables S8 and S9). As shown in Fig. S22 and S23, the device exhibits remarkable performance with a high V_OC of 0.890 V, even with a thicker active layer, and still achieves a PCE of 18.28% at a thickness of 446 nm. The external quantum efficiency (EQE) measurements (Fig. 4c) further corroborated the J–V results, showing a broadened and intensified photoresponse across the spectrum for the ternary device. The integrated J_cal values from EQE spectra agreed well with the measured J_SC values, with mismatches below 5%.

Fig. 4 (a) J–V curves of the binary and ternary OSCs. (b) Histograms of the PCE values of the binary and ternary OSCs. (c) External quantum efficiency (EQE) spectra. (d) The J–V characteristics of devices measured in the dark. Photoluminescence spectra at different excitation wavelengths: (e) 450 nm and (f) 730 nm. (g) The density-of-states (DoS) distribution. (h) Nyquist plots. (i) Mott–Schottky plots.

Table 2 Optimal binary and ternary devices under AM1.5G, 100 mW cm⁻² illumination

Photoactive layer	V OC [V]	J SC/Jcal.EQE [mA cm^−2]	FF [%]	PCE (PCEave^a) [%]
a Average values were obtained from 15 devices. b Certified efficiency of the PM6:L8-BO:BTP-FClO ternary device.
PM6:BTP-FClO	0.990 (0.985 ± 0.003)	21.54/20.34 (21.02 ± 0.06)	70.93 (70.41 ± 0.14)	15.13 (14.92 ± 0.05)
PM6:L8-BO	0.890 (0.886 ± 0.001)	27.56/26.33 (27.32 ± 0.06)	77.10 (76.92 ± 0.06)	18.91 (18.70 ± 0.05)
PM6:L8-BO:BTP-FClO	0.916 (0.908 ± 0.002)	28.02/27.08 (27.95 ± 0.04)	79.72 (79.40 ± 0.08)	20.46 (20.17 ± 0.05)
PM6:L8-BO:BTP-FClO^b	0.901	28.16	79.27	20.12

---

To elucidate the underlying mechanism of the enhanced device performance, we further examined charge recombination and exciton behavior using dark current and photoluminescence measurements. As shown in Fig. 4d, the PM6:L8-BO:BTP-FClO ternary device exhibits a lower dark current under forward bias than the two binary systems, indicating reduced parasitic resistance and superior rectification. More importantly, in the low bias region, the dark current of the ternary device is markedly suppressed, suggesting effective inhibition of non-radiative recombination losses. When excited with 450 nm (Fig. 4e), the PM6:BTP-FClO blend displays higher photoluminescence intensity than neat PM6. This is attributed to the strong aggregation tendency of BTP-FClO itself, which leads to the formation of large and dense aggregates in the blend film. These aggregates absorb energy from PM6 and re-emit intense fluorescence, resulting in extremely low J_SC and FF in the binary device. Under 730 nm excitation (Fig. 4f), the quenching efficiency of the PM6:L8-BO:BTP-FClO blend is further improved to 97.18%, exceeding that of the binary devices. This indicates that introducing a small amount of BTP-FClO does not compromise the original efficient charge-generation capability of L8-BO, but rather, through its interfacial modulation, further optimizes the nanoscale morphology of the acceptor phase.

To investigate the reasons for the enhanced exciton dissociation, we analyzed trap states and charge recombination. The PM6:BTP-FClO and PM6:L8-BO binary devices exhibited relatively high values of energetic disorder (σ) of 14.9 meV and 14.1 meV, respectively, whereas the PM6:L8-BO:BTP-FClO ternary blend achieved a reduced energetic disorder of 13.8 meV (Fig. 4g). As illustrated in Fig. 4h, Fig. S24 and S25, and Table S10, the Nyquist plots of all devices exhibit a single semicircular arc in the high-frequency region, dominated by the charge transfer resistance (R_ct), and a linear slope in the low-frequency region, corresponding to the series resistance (R_s). The ternary device, composed of PM6:L8-BO:BTP-FClO, exhibits the lowest charge transfer resistance (R_ct = 792 Ω) and series resistance (R_s = 82 Ω) among all devices, indicating improved charge transport and reduced ohmic losses. Furthermore, its highest capacitance (CPE-T = 6.152 nF) and near-ideal capacitive behavior (CPE-P = 0.971) suggest effective trap passivation and a reduced density of defect states, which are crucial for suppressing non-radiative recombination losses. As illustrated in Fig. 4i, the built-in potential (V_bi) is the primary driving force for the separation and transport of photogenerated charges. A higher V_bi implies a stronger built-in electric field within the device. The ternary device exhibits the highest V_bi (0.91 V), which provides a fundamental physical basis for achieving a higher V_OC.

To investigate the impact of BTP-FClO on charge recombination in PM6:L8-BO-based devices, we examined the relationship between J_SC and V_OC as a function of light intensity (Fig. 5a and b). The relationship between V_OC and light intensity (P_light) follows V_OC ∝ nkT/q ln(P_light), where an ideality factor (n) closer to 1 indicates that bimolecular recombination is the dominant loss mechanism, while larger slopes suggest that bulk-traps are dominating the non-geminate recombination.^51–53 PM6:L8-BO:BTP-FClO exhibits the smallest slope (1.12kT/q). Efficient charge extraction is indicated by the relationship J_SC ∝ P^α_light, where an α value close to 1 suggests efficient charge extraction prior to recombination. The ternary device exhibits the highest α value (0.999), which is extremely close to the ideal value of 1.^54,55 Moreover, the highly balanced hole and electron mobility further supports that the near-unity α value arises from the combined effects of suppressed bimolecular recombination and balanced charge transport.^56,57Fig. 5c establishes the theoretical relationship between the ideal FF under the influence of V_OC. The PM6:L8-BO:BTP-FClO blend closely aligns with the FF_max −0.06 curve, indicating minimal losses from non-ideal factors, such as series resistance or imbalanced charge transport.^58–60 This demonstrates exceptional agreement between the theoretical model and experimental results. The introduction of BTP-FClO optimizes the active-layer morphology, thereby enabling an energy conversion system with reduced recombination losses and enhanced charge-extraction efficiency.

Fig. 5 (a) V_OC dependence on the light intensity. (b) J_SC dependence on the light intensity. (c) FF values as a function of V_OC for this work and previously reported devices. (d) Deep-level transient spectroscopy (DLTS) analysis. (e) Charge carrier density as a function of the delay time of the devices. (f) TPC and TPV results of the binary and ternary-based OSCs. (g) The temperature varied the thermally annealed film's UV-vis spectra and deviation metrics of L8-BO, L8-BO:BTP-FClO, and BTP-FClO. Stability of the binary and ternary devices: (h) Normalized efficiency evolution under continuous LED illumination. (i) Thermal stability under an inert atmosphere (N₂) at 60 °C.

Furthermore, based on deep-level transient spectroscopy (DLTS) analysis, the trap state signal peak intensity of the PM6:L8-BO:BTP-FClO ternary device is 2.69 × 10¹⁶ cm⁻³, lower than that of the PM6:L8-BO and PM6:BTP-FClO binary devices (2.99 × 10¹⁶ cm⁻³ and 3.20 × 10¹⁶ cm⁻³, respectively) as shown in Fig. 5d. This indicates a notable reduction in the deep-level trap density in the ternary blend system, corroborating the DOS results. As presented in Fig. 5e and Table S11, the bimolecular recombination coefficient (β = 0.95 × 10⁻¹² cm³ s⁻¹) of the PM6:L8-BO:BTP-FClO ternary device is lower than that of the two binary systems (PM6:BTP-FClO: 4.90 × 10⁻¹² cm³ s⁻¹; PM6:L8-BO: 2.43 × 10⁻¹² cm³ s⁻¹). A lower β value indicates a substantially reduced probability that free electrons and holes encounter each other and undergo bimolecular recombination.^61–63 Therefore, under the same illumination conditions, the ternary device demonstrates the highest initial carrier concentration (n₀ = 8.25 × 10¹⁶ cm⁻³), which directly confirms more efficient charge transport and reduced recombination losses in the ternary device. This demonstrates that the introduction of BTP-FClO effectively reduces trap state density and lowers the trap energy levels by optimizing the interface structure and suppressing defect formation, thereby underpinning the exceptional photovoltaic performance of the ternary device.

To directly investigate the impact of the aforementioned advantages on charge behavior from a kinetic perspective, we analyzed carrier extraction time and recombination lifetime using transient photocurrent (TPC) and transient photovoltage (TPV) measurements, as shown in Fig. 5f and Fig. S26. TPC measures the decay process of the photogenerated current, and its decay time constant (τ_TPC) reflects the carrier extraction time.^64–66 The decay time constants of PM6:BTP-FClO, PM6:L8-BO, and PM6:L8-BO:BTP-FClO devices are 0.31, 0.26, and 0.24 µs, respectively. A faster extraction speed indicates that carriers in the ternary system encounter fewer recombination opportunities during transport, resulting in improved charge extraction efficiency and directly contributing to higher J_SC and FF. TPV measures the decay of the photogenerated voltage, and its decay time constant (τ_TPV) directly represents the free-carrier lifetime. The ternary device exhibits the longest τ (6.93 µs), higher than those of the two binary systems (PM6:L8-BO: 5.35 µs and PM6:BTP-FClO: 3.32 µs). A longer lifetime means that photogenerated electrons and holes remain separated for longer after their initial separation, effectively suppressing non-radiative recombination.^67–69 This provides direct kinetic evidence for the device achieving a higher V_OC.

After verifying the exceptional photoelectric conversion efficiency of the ternary device, we further evaluated its operational and storage stability, which are critical metrics for assessing the practical application potential of such devices, as shown in Fig. 5g–i and Fig. S27, S28. Under continuous LED exposure, the ternary device's stability advantage was equally pronounced. After more than 1200 h of continuous illumination, its normalized efficiency remained above 90% of its initial value, with an exceptionally gradual decay. Conversely, the PM6:L8-BO binary device showed a more rapid and significant degradation under the same conditions. The PM6:L8-BO binary device exhibited inferior thermal stability, with its efficiency rapidly decaying to 80% of the initial value (T₈₀) within 72 hours, followed by a swift decline thereafter. In contrast, the ternary device's efficiency degradation was remarkably slow, reaching T₈₀ only after 448 hours. This dramatic enhancement in stability is directly attributed to the elevated glass transition temperature (T_g) of the active layer. The incorporation of the high-T_g acceptor BTP-FClO (128 °C) into the L8-BO (105 °C) host notably increased the T_g to 120 °C. A higher T_g signifies a more rigid and stable amorphous phase, which effectively suppresses the detrimental molecular diffusion and reorganization of the acceptors under continuous thermal stress (60 °C). The incorporation of BTP-FClO optimized the initial morphology by serving as an interfacial compatibilizer to enhance efficiency and as a powerful morphology stabilizer. By strengthening intermolecular interactions, it anchored the optimized morphology, evidently and multiplicatively improving the long-term stability of the device under both thermal and light stress.

Inspired by the properties of BTP-FClO, we further evaluated its versatility as a third component in a set of high-performance donor–acceptor combinations (Fig. 6a). Based on contact angle measurements and surface free energy calculations in Fig. S29 and Table S12, the wetting coefficient values of BTP-FCIO in the three systems (Y6, L8-BO-X, and BTP-eC9) are 1.20, 2.69, and 1.24, respectively, indicating that BTP-FCIO preferentially resides within the acceptor phase. As illustrated in Fig. S30 and S31, the binary blends exhibit nucleation onset times of approximately 138 ms. Remarkably, after incorporating BTP-FCIO, all ternary blends display substantially delayed nucleation (∼200 ms). As shown in Fig. 6b and Table 3, the introduction of BTP-FClO into three binary systems considerably enhanced the photovoltaic performance. Specifically, the champion PCEs of the PM6:Y6:BTP-FClO, PM6:BTP-eC9:BTP-FClO, and PM6:L8-BO-X:BTP-FClO devices increased to 18.45%, 19.37%, and 19.34%, respectively. This improvement is primarily attributed to the elevated LUMO energy level of BTP-FClO, which led to a gain in open-circuit voltage. In the systems based on Y6, BTP-eC9, and L8-BO-X, the champion V_OC values increased from 0.842 V to 0.858 V, 0.849 V to 0.871 V, and 0.888 V to 0.894 V, respectively. As illustrated in Fig. 6c, the EQE spectra of the three systems indicate that incorporating BTP-FClO enhanced photoresponse, thereby significantly improving the J_SC. The integrated current densities (J_cal,EQE) for the three devices were 27.17, 27.66, and 26.71 mA cm⁻², respectively, and showed good agreement with the measured J_SC values, with mismatches all below 5%. These results demonstrate that BTP-FClO is an efficient and versatile third component with considerable potential for developing high-performance ternary OSCs.

Fig. 6 (a) Chemical structures of Y6, BTP-eC9, and L8-BO-X. (b) J–V curves and (c) external quantum efficiency (EQE) of these three OSC systems.

Table 3 Optimal binary and ternary devices under AM1.5G, 100 mW cm⁻² illumination

Photoactive layer	V OC [V]	J SC/Jcal.EQE [mA cm^−2]	FF [%]	PCE (PCEave^a) [%]
a Average values were obtained from 10 devices.
PM6:Y6	0.842	26.71/25.96	76.76	17.25 (17.13 ± 0.04)
PM6:Y6:BTP-FClO	0.858	27.71/27.17	77.60	18.45 (18.26 ± 0.07)
PM6:BTP-eC9	0.849	27.11/26.33	78.07	17.96 (17.79 ± 0.06)
PM6:BTP-eC9:BTP-FClO	0.871	28.11/27.66	79.07	19.37 (19.20 ± 0.05)
PM6:L8-BO-X	0.888	26.18/25.41	77.89	18.13 (18.02 ± 0.04)
PM6:L8-BO-X:BTP-FClO	0.894	27.65/26.71	78.21	19.34 (19.17 ± 0.05)

---

Conclusions

In summary, we introduced the end-group-asymmetric NFA BTP-FClO as a third component into the PM6:L8-BO blend. By slowing the crystallization kinetics, BTP-FClO extends the nucleation onset of the acceptor phase from 132 ms to 199 ms, thereby acting as an effective compatibilizer at the donor–acceptor interface. The incorporation of BTP-FClO enabled the formation of a refined, continuous interpenetrating network with tighter π–π stacking (3.27 Å) and enhanced molecular ordering. These morphological advantages facilitated balanced hole and electron transport (μ_e/μ_h = 1.06) and effectively suppressed charge recombination. Consequently, ternary devices achieved a PCE of 20.46% (certified: 20.12%) and exhibited evidently improved operational stability under both thermal and illumination stress, substantially outperforming their binary counterparts. Remarkably, a PCE of 18.28% was preserved at a thick active layer of 446 nm, demonstrating exceptional thickness insensitivity and compatibility with large-scale processing. These findings elucidate a molecular design strategy that leverages tailored intermolecular interactions to deliberately slow nucleation kinetics, thereby optimizing blend morphology for simultaneous improvements in efficiency and stability, with broad applicability to other Y-series acceptors.

Author contributions

A. L. and C. L. contributed equally to this work. A. L. wrote the original manuscript draft, fabricated the devices, conducted the measurements, and performed in-depth data analysis. C. L. designed the materials and analyzed the data. S. C., B. K., Y. C., and K. C. conducted the GIWAXS measurements and data analysis and provided analysis tools. J. Z., J. Z., G. Z., L. T., X. L., and W. W. performed the data analysis of TPC, TPV, and DLTS measurements. B. Z. and H. Y. designed the materials and revised the manuscript. Z. K. supervised the experiments and provided resources, administration, and funding. All authors read, corrected, and approved the manuscript.

Conflicts of interest

There are no conflicts to declare.

Data availability

The data supporting this article have been included as part of the supplementary information (SI).

Supplementary information: raw data for absorbance, PL, GIWAXS, contact angle, theoretical calculations, ultraviolet photoelectron spectroscopy, in situ 2D UV/Vis absorption, and photo-CELIV measurements. See DOI: https://doi.org/10.1039/d6ee01250g.

Acknowledgements

The authors acknowledge the financial support from the National Natural Science Foundation of China (62275057), the Guangxi Natural Science Foundation (2023GXNSFFA026004 and 2022GXNSFDA035066), and the Innovation Project of Guangxi Graduate Education (YCBZ2024034). Portions of this research were carried out at the 3C SAXS-I and 9A U-SAXS beamlines of the Pohang Accelerator Laboratory (PLS-II), Republic of Korea.

References

1. R. Ma, Z. Luo, Y. Zhang, L. Zhan, T. Jia, P. Cheng, C. Yan, Q. Fan, S. Liu, L. Ye, G. Zhang, X. Xu, W. Gao, Y. Wu, J. Wu, Y. Li, Y. Liu, F. Liu, J. Song, H. Chen, W. Chen, X. Zhang, Y. Liu, J. Yuan, Q. Liu, Z. Kan, H. Yin, X. Li, Y. Ma, D. Deng, L. Zhu, Y. Huo, B. Fan, H. Fu, X. Liao, H. Hu, C. Li, R. Yu, H. Hu, Z. Yao, Y. Cai, D. Qian, Y. Cui, H. Yao, B. Xu, B. Kan, K. Gao, C. Duan, X. Hu and H. Sun, Sci. China Mater., 2025, 68, 1689–1701.
2. B. Zou, W. Wu, T. A. Dela Peña, R. Ma, Y. Luo, Y. Hai, X. Xie, M. Li, Z. Luo, J. Wu, C. Yang, G. Li and H. Yan, Nano-Micro Lett., 2023, 16, 30.
3. S. Karuthedath, S. H. K. Paleti, A. Sharma, H. Yin, C. S. P. De Castro, S. Chen, H. Xu, N. Alshehri, N. Ramos, J. I. Khan, J. Martin, G. Li, F. Laquai, D. Baran and J. Gorenflot, Adv. Energy Mater., 2023, 13, 2203464.
4. Y. Zheng, J. Zhao, H. Liang, Z. Zhao and Z. Kan, Adv. Sci., 2023, 10, 2302460.
5. J. Zhao, S. Chung, H. Li, Z. Zhao, C. Zhu, J. Yin, K. Cho and Z. Kan, Adv. Funct. Mater., 2023, 33, 2307355.
6. B. Zou, H. M. Ng, Z. Li, Y. Wang, Q. Wang, D. Chen, Z. Yao, H. Li, C. Li, X. Zeng, W. Liu, J. E. Halpert, H. Hu, C. Duan, Z. Zhu, T. Wu, W. Wong, Z. Zhang, H. Yan and H. Yu, Adv. Sci., 2025, 12, e12690.
7. Z. Ling, J. Wu, J. P. Jurado, C. E. Petoukhoff, S. Y. Jeong, D. Naphade, M. Babics, X. Chang, H. Faber, S. Doukas, E. Lidorikis, M. I. Nugraha, M. He, M. Alqurashi, Y. Lin, X. Sun, H. Hu, H. Y. Woo, S. De Wolf, L. Tsetseris, F. Laquai, D. Yu, E. Wang and T. D. Anthopoulos, Mater. Sci. Eng., R, 2025, 163, 100922.
8. C. Li, Y. Cai, P. Hu, T. Liu, L. Zhu, R. Zeng, F. Han, M. Zhang, M. Zhang, J. Lv, Y. Ma, D. Han, M. Zhang, Q. Lin, J. Xu, N. Yu, J. Qiao, J. Wang, X. Zhang, J. Xia, Z. Tang, L. Ye, X. Li, Z. Xu, X. Hao, Q. Peng, F. Liu, L. Guo and H. Huang, Nat. Mater., 2025, 24, 1626–1634.
9. L. Zhu, M. Zhang, G. Zhou, Z. Wang, W. Zhong, J. Zhuang, Z. Zhou, X. Gao, L. Kan, B. Hao, F. Han, R. Zeng, X. Xue, S. Xu, H. Jing, B. Xiao, H. Zhu, Y. Zhang and F. Liu, Joule, 2024, 8, 3153–3168.
10. L. Wang, C. Chen, Z. Gan, J. Cheng, Y. Sun, J. Zhou, W. Xia, D. Liu, W. Li and T. Wang, Adv. Mater., 2025, 37, 2419923.
11. H. Chen, Y. Huang, R. Zhang, H. Mou, J. Ding, J. Zhou, Z. Wang, H. Li, W. Chen, J. Zhu, Q. Cheng, H. Gu, X. Wu, T. Zhang, Y. Wang, H. Zhu, Z. Xie, F. Gao, Y. Li and Y. Li, Nat. Mater., 2025, 24, 444–453.
12. R. Ma, B. Zou, Y. Hai, Y. Luo, Z. Luo, J. Wu, H. Yan and G. Li, Adv. Mater., 2025, 37, 2500861.
13. H. Yao, L. Ye, J. Hou, B. Jang, G. Han, Y. Cui, G. M. Su, C. Wang, B. Gao, R. Yu, H. Zhang, Y. Yi, H. Y. Woo, H. Ade and J. Hou, Adv. Mater., 2017, 29, 1700254.
14. B. Zou, H. M. Ng, H. Yu, P. Ding, J. Yao, D. Chen, S. H. Pun, H. Hu, K. Ding, R. Ma, M. Qammar, W. Liu, W. Wu, J. Y. L. Lai, C. Zhao, M. Pan, L. Guo, J. E. Halpert, H. Ade, G. Li and H. Yan, Adv. Mater., 2024, 36, 2405404.
15. Z. Zhao, S. Chung, Y. Y. Kim, M. Jeong, X. Li, J. Zhao, C. Zhu, S. Karuthedath, Y. Zhong, K. Cho and Z. Kan, Energy Environ. Sci., 2024, 17, 5666–5678.
16. C. Zhu, S. Chung, J. Zhao, Y. Sun, B. Zhao, Z. Zhao, S. Kim, K. Cho and Z. Kan, Adv. Sci., 2023, 10, 2303150.
17. W. Wu, B. Zou, R. Ma, J. Yao, C. Li, Z. Luo, B. Xie, M. Qammar, T. A. Dela Peña, M. Li, J. Wu, C. Yang, Q. Fan, W. Ma, G. Li and H. Yan, Small, 2024, 20, 2402793.
18. Y. Zhang, J. Deng, S. You, X. Huang, J. Liu, Y. Cheng, B. Huang, X. Chen, Z. Sun, C. Yang, Q. Cao, F. Wu and L. Chen, Adv. Funct. Mater., 2024, 34, 2308151.
19. H. Zhang, Y. Chai, J. Zhao, E. Zhu, Y. Chen, L. Liu and D. M. Guldi, Adv. Energy Mater., 2025, 15, 2500332.
20. A. Liang, Y. Sun, S. Chung, J. Shin, K. Sun, C. Zhu, J. Zhao, Z. Zhao, Y. Zhong, G. Zhang, K. Cho and Z. Kan, Nano-Micro Lett., 2025, 17, 72.
21. B. Zhao, S. Chung, M. Zhang, W. Wei, C. Zhu, C. Deng, K. Cho and Z. Kan, Adv. Funct. Mater., 2024, 34, 2309832.
22. J. Guo, S. Qin, J. Zhang, C. Zhu, X. Xia, Y. Gong, T. Liang, Y. Zeng, G. Han, H. Zhuo, Y. Li, L. Meng, Y. Yi, J. Chen, X. Li, B. Qiu and Y. Li, Nat. Commun., 2025, 16, 1503.
23. L. Wang, C. Chen, Y. Fu, C. Guo, D. Li, J. Cheng, W. Sun, Z. Gan, Y. Sun, B. Zhou, C. Liu, D. Liu, W. Li and T. Wang, Nat. Energy, 2024, 9, 208–218.
24. C. Li, G. Lu, H. S. Ryu, X. Sun, H. Y. Woo and Y. Sun, ACS Appl. Mater. Interfaces, 2022, 14, 43207–43214.
25. Q. Wei, S. Liang, W. Liu, Y. Hu, B. Qiu, J. Ren, J. Yuan, F. Huang, Y. Zou and Y. Li, ACS Energy Lett., 2022, 7, 2373–2381.
26. H. M. Ng, B. Zou, A. Sergeev, Y. Fu, P. F. Chan, Z. Yao, Q. Wang, Z. Li, C.-J. Su, U.-S. Jeng, X. Hu, G. Li, X. Lu, K. S. Wong, Z.-G. Zhang, Y. Chen, W.-Y. Wong, H. Yu and H. Yan, Energy Environ. Sci., 2025, 18, 6587–6596.
27. X. Liu, Y. Yan, Y. Yao and Z. Liang, Adv. Funct. Mater., 2018, 28, 1802004.
28. Y. Xie, T. Li, J. Guo, P. Bi, X. Xue, H. S. Ryu, Y. Cai, J. Min, L. Huo, X. Hao, H. Y. Woo, X. Zhan and Y. Sun, ACS Energy Lett., 2019, 4, 1196–1203.
29. X. Duan, W. Song, J. Qiao, X. Li, Y. Cai, H. Wu, J. Zhang, X. Hao, Z. Tang, Z. Ge, F. Huang and Y. Sun, Energy Environ. Sci., 2022, 15, 1563–1572.
30. Y. Li, Y. Cai, Y. Xie, J. Song, H. Wu, Z. Tang, J. Zhang, F. Huang and Y. Sun, Energy Environ. Sci., 2021, 14, 5009–5016.
31. K. Chong, X. Xu, H. Meng, J. Xue, L. Yu, W. Ma and Q. Peng, Adv. Mater., 2022, 34, 2109516.
32. Q. An, J. Wang, X. Ma, J. Gao, Z. Hu, L. Bin, H. Sun, X. Gao, X. Zhang and F. Zhang, Environ. Sci., 2020, 13, 5039.
33. J. Fu, Q. Yang, P. Huang, S. Chung, K. Cho, Z. Kan, H. Liu, X. Lu, Y. Lang, H. Lai, F. He, P. W. K. Fong, S. Lu, Y. Yang, Z. Xiao and G. Li, Nat. Commun., 2024, 15, 1830.
34. J. Zhang, J. Zhong, S. Chung, L. Bai, Z. Zhao, L. Tan, Y. Liu, M. Zhang, S. Lyu, J. Lee, L. Li, K. Cho and Z. Kan, Chem. Eng. J., 2025, 522, 167311.
35. B. Zou, A. Liang, P. Ding, J. Yao, X. Zeng, H. Li, R. Ma, C. Li, W. Wu, D. Chen, M. Qammar, H. Yu, J. Yi, L. Guo, S. H. Pun, J. E. Halpert, G. Li, Z. Kan and H. Yan, Angew. Chem., 2025, 137, e202415332.
36. C. Zhu, A. Liang, S. Chung, G. Zhang, S. Yun, K. H. Cho, Y. Tang, K. Cho, Z. Zhao and Z. Kan, Chin. J. Chem., 2026, 44, 1203–1211.
37. C. Gao, Y. Wang, H. Tian, K. Sun, C. Zhao, J. Che, S. Qiu, W. Chen, C. Huang, Z. Wang, H. Hu, P. Han, S. Li, Z. Luo and G. Zhang, Adv. Mater., 2025, e17576.
38. Q. Fan, W. Su, Y. Wang, B. Guo, Y. Jiang, X. Guo, F. Liu, T. P. Russell, M. Zhang and Y. Li, Sci. China: Chem., 2018, 61, 531–537.
39. J. Fu, Q. Yang, P. Huang, S. Chung, K. Cho, Z. Kan, H. Liu, X. Lu, Y. Lang, H. Lai, F. He, P. W. K. Fong, S. Lu, Y. Yang, Z. Xiao and G. Li, Nat. Commun., 2024, 15, 1830.
40. C. Li, A. Liang, S. Qiu, G. H. M. Ng, A. Sergeev, J. Y. L. Lai, B. Zou, X. Zeng, J. Yi, H. Li, J. Guo, W. Liu, J. Yao, K. S. Wong, Z. Kan, S. H. Pun and H. Yan, Adv. Funct. Mater., 2026, e31879.
41. J.-S. Huang, T. Goh, X. Li, M. Y. Sfeir, E. A. Bielinski, S. Tomasulo, M. L. Lee, N. Hazari and A. D. Taylor, Nat. Photonics, 2013, 7, 479–485.
42. R. Yu, S. Zhang, H. Yao, B. Guo, S. Li, H. Zhang, M. Zhang and J. Hou, Adv. Mater., 2017, 29, 1700437.
43. W. Gao, H. Fu, Y. Li, F. Lin, R. Sun, Z. Wu, X. Wu, C. Zhong, J. Min, J. Luo, H. Y. Woo, Z. Zhu and A. K.-Y. Jen, Adv. Energy Mater., 2021, 11, 2003177.
44. B. Zhao, X. Huang, S. Chung, M. Zhang, Y. Zhong, A. Liang, Z. Zhao, C. Zhu, J. Zhao, S. Kim, J. Kim, M. Wang, S. Chen, K. Cho, Y. Wang and Z. Kan, eScience, 2024, 100305.
45. Z. Zhao, J. Zhao, S. Chung, K. Cho, W. Xu and Z. Kan, ACS Mater. Lett., 2023, 5, 1718–1726.
46. W. Liang, S. Zhu, K. Sun, J. Hai, Y. Cui, C. Gao, W. Li, Z. Wu, G. Zhang and H. Hu, Adv. Funct. Mater., 2025, 35, 2415499.
47. Y. Sun, S. Chung, X. Huang, K. Cho and Z. Kan, Chem. Eng. J., 2023, 470, 144186.
48. S. Honda, H. Ohkita, H. Benten and S. Ito, Adv. Energy Mater., 2011, 1, 588–598.
49. J. Lv, Q. Yang, W. Deng, H. Chen, M. Kumar, F. Zhao, S. Lu, H. Hu and Z. Kan, Chem. Eng. J., 2023, 465, 142822.
50. V. Chellappan, G. M. Ng, M. J. Tan, W.-P. Goh and F. Zhu, Appl. Phys. Lett., 2009, 95, 263305.
51. C. M. Proctor, M. Kuik and T.-Q. Nguyen, Prog. Polym. Sci., 2013, 38, 1941–1960.
52. J. Vollbrecht and V. V. Brus, Org. Electron., 2020, 86, 105905.
53. G. Akhtanova, H. P. Parkhomenko, J. Vollbrecht, A. I. Mostovyi, N. Schopp and V. Brus, Org. Electron., 2025, 137, 107183.
54. X. Wu, X. Zhang, J. Zhang, Y. Wu, J. Li, X. Kong, Z. Li, X. Zhang, X. Li, A. Li, G. Yang and C. Sun, Adv. Funct. Mater., 2024, 34, 2405168.
55. J. Lv, Q. Yang, W. Deng, H. Chen, M. Kumar, F. Zhao, S. Lu, H. Hu and Z. Kan, Chem. Eng. J., 2023, 465, 142822.
56. U. Würfel, D. Neher, A. Spies and S. Albrecht, Nat. Commun., 2015, 6, 6951.
57. J. Vollbrecht, V. V. Brus, S. Ko, J. Lee, A. Karki, D. X. Cao, K. Cho, G. C. Bazan and T. Nguyen, Adv. Energy Mater., 2019, 9, 1901438.
58. C. Xu, J. Yang, S. Gámez-Valenzuela, J.-W. Lee, J. Che, P. Chen, G. Zhang, D. Hu, Y. Wang, J. Lv, Z. Zhong, X. Chen, G. Zhang, F. Zhao, B. J. Kim, X. Guo and B. Liu, Energy Environ. Sci., 2025, 18, 5913–5925.
59. K. Xing, D. Cai, D. Wang, J.-Y. Wang, C. Tang, Y. Ma and Q. Zheng, Natl. Sci. Rev., 2025, 12, nwaf089.
60. B. Zhao, L. Zhu, S. Xiong, J. Yu, X. Wang, J. Zhao, L. Tan, J. Zhang, J. Zhong, L. Kan, X. Wan, K. Jiang, H. Li, Z. Ma, Y. Liu, H. Zhu, Z. Kan, F. Liu, Z. Sun, J. Chu and Q. Bao, Adv. Energy Mater., 2025, 15, e04947.
61. C. G. Shuttle, B. O’Regan, A. M. Ballantyne, J. Nelson, D. D. C. Bradley and J. R. Durrant, Phys. Rev. B:Condens. Matter Mater. Phys., 2008, 78, 113201.
62. N. Gasparini, A. Katsouras, M. I. Prodromidis, A. Avgeropoulos, D. Baran, M. Salvador, S. Fladischer, E. Spiecker, C. L. Chochos, T. Ameri and C. J. Brabec, Adv. Funct. Mater., 2015, 25, 4898–4907.
63. J. Min, Y. N. Luponosov, N. Gasparini, M. Richter, A. V. Bakirov, M. A. Shcherbina, S. N. Chvalun, L. Grodd, S. Grigorian, T. Ameri, S. A. Ponomarenko and C. J. Brabec, Adv. Energy Mater., 2015, 5, 1500386.
64. X. Song, K. Zhang, R. Guo, K. Sun, Z. Zhou, S. Huang, L. Huber, M. Reus, J. Zhou, M. Schwartzkopf, S. V. Roth, W. Liu, Y. Liu, W. Zhu and P. Müller-Buschbaum, Adv. Mater., 2022, 34, 2200907.
65. Z. Li, F. Gao, N. C. Greenham and C. R. McNeill, Adv. Funct. Mater., 2011, 21, 1419–1431.
66. Z. Zhao, S. Chung, L. Tan, J. Zhao, Y. Liu, X. Li, L. Bai, H. Lee, M. Jeong, K. Cho and Z. Kan, Energy Environ. Sci., 2025, 18, 2791–2803.
67. X. Li, Y. Cho, S. Chung, J. Zhong, M. Zhang, A. Liang, Z. Zhao, S. Karuthedath, K. Cho and Z. Kan, Adv. Funct. Mater., 2026, 36(8), e21694.
68. Z. Zhao, Y. Liang, H. Tang, J. Zhong, C. Cao, Y. Qin, H. Li, L. Bai, W. Pang, G. Chen, S. Karuthedath and Z. Kan, J. Energy Chem., 2026, 116, 251–261.
69. L. Tan, Z. Zhao, J. Zhang, H. Li, M. Zhang, J. Zhao, Y. Liu, L. Bai, T. Jia and Z. Kan, Mater. Sci. Eng., R, 2026, 169, 101205.

---

Footnote

† These authors contributed equally to this work.

---