Aggregation‐Induced Emission Molecular Design for Mitigating Non‐Radiative Energy Loss in Organic Solar Cells

ABSTRACT

Non‐radiative energy loss remains a critical bottleneck limiting the open‐circuit voltage (V _OC) and efficiency of organic solar cells (OSCs). Here, we introduce a molecular design strategy that leverages aggregation‐induced emission (AIE) to suppress aggregation‐caused quenching and enhance solid‐state photoluminescence quantum yield (PLQY), thereby mitigating non‐radiative recombination. A prototypical AIE motif, tetraphenylethylene, was incorporated into the terminal group of a Y‐series non‐fullerene acceptor to yield dTPE, which exhibits distinct AIE characteristics not previously observed in high‐performance Y‐series acceptors. Photoluminescence studies reveal that dTPE achieves a threefold enhancement in PLQY compared to L8BO‐C4 in the film, leading to an electroluminescence external quantum efficiency more than an order of magnitude higher than that of D18:L8BO‐C4. Consequently, the binary D18:dTPE device achieves a remarkably low non‐radiative recombination loss of 0.130 eV. When incorporated as a guest into D18:L8BO‐C4 blends, dTPE enables a non‐radiative voltage loss of only 0.190 eV and an unprecedented V _OC of 0.93 V, yielding an efficiency of 20.5%. To our knowledge, this represents the highest V _OC reported for OSCs with efficiencies above 20%. This work establishes AIE molecular design as an effective pathway to overcome intrinsic limitations of Y‐series acceptors and provides guiding principles for mitigating non‐radiative energy loss in next‐generation OSCs.

Aggregation‐induced emission strategy is first used to overcome aggregation‐caused quenching in Y‐series acceptors. By integrating tetraphenylethylene, the dTPE exhibits three‐fold PLQY enhancement and significantly suppressed non‐radiative energy loss. An ultra‐low E _loss of 0.130 eV and a record V _OC of 0.93 V are achieved, resulting in a PCE over 20.5%. This work establishes AIE design as a guiding principle for high‐performance organic photovoltaics.

1. Introduction

Organic solar cells (OSCs) have made remarkable progress in the past decade, with power conversion efficiencies (PCEs) now exceeding 20% [1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12]. Nevertheless, further advancement toward the efficiency range of perovskite and silicon photovoltaics is hampered by large non‐radiative energy losses (ΔE _nr), which severely suppress the open‐circuit voltage (V _OC) and limit device performance [13, 14, 15, 16, 17, 18, 19, 20]. One of the fundamental challenges lies in the molecular design of state‐of‐the‐art Y‐series non‐fullerene acceptors (NFAs) [21, 22, 23, 24, 25, 26, 27]. Their rigid and planar donor–acceptor backbones favor charge transfer but simultaneously suppress radiative decay. As a result, nearly all high‐performing NFAs are weakly luminescent in the solid state (Figure 1a, left), constrained by aggregation‐caused quenching (ACQ) behavior that enforces persistent ΔE _nr losses [28, 29]. Overcoming this photophysical bottleneck is widely regarded as a critical frontier for breaking the efficiency plateau of OSCs [30, 31, 32, 33].

FIGURE 1.

(a) Molecular design and chemical structures of L8BO‐C4, tetraphenylethylene, 9,9′‐spirobi[fluorene], dTPE, and dSpiro. Normalized absorption spectra of these materials (b) in solutions and (c) in films. (d) Energy level diagram of the materials in this work. (e) PL spectra of dTPE in the mixture of tetrahydrofuran (THF) and water. (f) Plots of the relative emission intensity of L8BO‐C4 and dTPE versus water fraction. (g) PLQY of L8BO‐C4, dTPE, and dSpiro neat films.

One widely explored approach to reducing ΔE _nr has been the incorporation of ternary guest components, which often yield improved photoluminescence quantum yield (PLQY) and electroluminescence quantum efficiency (EQE_EL) [34, 35, 36]. A seminal study by Gao and co‐workers established guiding principles for the design of guest materials [37]. The guest‐based binary blend should exhibit a higher EQE_EL than the host system, thereby facilitating thermal population of radiative states and suppressing non‐radiative decay. Consistent with these principles, many successful ternary systems employ luminescent guest acceptors. For example, Sun and co‐workers introduced a triphenylamine‐functionalized NFA (Z‐Tri) that suppressed non‐radiative losses and boosted EQE_EL by more than an order of magnitude compared with PM6:L8‐BO [38]. We demonstrated oligomers with D‐A‐D‐A‐D and A‐D‐A‐D‐A configuration(named 5BDTBDD and 5BDDBDT) give higher EQE_EL of up to 0.05%, enable high‐performance ternary OSCs with low energy loss [34]. These studies highlight the importance of enhancing both PLQY and EQE_EL in acceptors as a direct route to suppress ΔE _nr. However, the intrinsic ACQ character of Y‐series cores remains unchanged, placing an inherent ceiling on such strategies.

A transformative solution is to invert this paradigm by harnessing aggregation‐induced emission (AIE). In contrast to ACQ systems, AIE luminogens are non‐emissive in dilute solution but become highly luminescent upon aggregation, making them ideally suited for the condensed morphology of OSC active layers [39, 40]. Embedding AIE motifs into NFAs thus offers a conceptually powerful and previously unexplored strategy to couple favorable electronic properties with enhanced solid‐state luminescence. In this study, we present a new design strategy that integrates AIE‐active motifs directly into Y‐series acceptors. By incorporating tetraphenylethylene (TPE) into the terminal groups, we developed the acceptor dTPE (Figure 1a, middle), along with a control analogue dSpiro based on spirobifluorene (Figure 1a, right), which features a rigid aromatic group of similar size but lacks AIE character. Strikingly, dTPE films exhibit pronounced AIE behavior with a PLQY three times higher than L8BO‐C4, a feature not previously observed in high‐performance Y‐series acceptors, whereas dSpiro retains conventional ACQ character. As a result, the guest binary of D18:dTPE achieved an EQE_EL more than an order of magnitude compared with the host binary of D18:L8BO‐C4 and a remarkably low ΔE _nr of 0.130 eV. When incorporated dTPE as a guest into D18:L8‐BO‐C4 blends, the enhanced PLQY translated into a remarkably low non‐radiative voltage loss of 0.190 eV and an unprecedented V _OC of 0.93 V, yielding a PCE of 20.5%. To our knowledge, this represents the highest V _OC reported for OSCs with efficiencies above 20%.

More than achieving another efficiency milestone, this work establishes a new conceptual framework for molecular design in OSCs. By transforming ACQ into AIE, we demonstrate a generalizable approach to mitigating non‐radiative recombination losses at the molecular level. These findings position AIE not as a niche phenomenon but as a guiding design principle for next‐generation organic semiconductors, pointing the way toward OSCs with both high efficiency and intrinsically low energy loss.

2. Results and Discussion

The synthetic routes for dTPE and dSpiro are straightforward, requiring only an additional Suzuki coupling step to obtain the IC‐TPE and IC‐Spiro terminal groups from commercially available bromine precursors. Detailed synthesis procedures and characterization, including NMR and mass spectra, are provided in Schemes S1 and S2 and Figures S19–S28. Both molecules show good solubility in common processing solvents such as chloroform and chlorobenzene, ensuring their suitability for device fabrication.

The optical absorption spectra of L8BO‐C4, dTPE, and dSpiro in solution and thin film are compared in Figure 1b,c, with key parameters summarized in Table S3. In chloroform solution, all three molecules display nearly identical absorption maxima around 727–728 nm, with extinction coefficients in the range of 1.2–1.8 × 10⁵ L·mol^{− 1}·cm^{− 1}, indicating that incorporation of TPE or Spiro units exerts minimal influence on the intrinsic electronic transitions of the molecular backbone. By contrast, notable differences arise in the solid state. Upon film formation, L8BO‐C4 exhibits a pronounced redshift of 74 nm, whereas dTPE and dSpiro show more moderate shifts of 52 and 53 nm, respectively, reflecting their distinct aggregation behaviors. The corresponding optical bandgaps (E _g) were determined to be 1.40 eV for L8BO‐C4, 1.46 eV for dTPE, and 1.44 eV for dSpiro. Charge transport was evaluated using electron‐only devices via the space‐charge limited current (SCLC) method (Figure S8). The electron mobilities (µ _e) decrease in the order L8BO‐C4 > dTPE > dSpiro, with values of 9.67 × 10^{− 4}, 5.28 × 10^{− 4}, and 2.87 × 10^{− 4} cm² V^{− 1} s^{− 1}, respectively. Cyclic voltammetry (Figure S3) revealed slight upshifts in both highest occupied molecular orbital (HOMO) and lowest unoccupied molecular orbital (LUMO) levels for dTPE (−5.56/−3.83 eV) and dSpiro (−5.47/−3.77 eV) compared to L8BO‐C4 (−5.64/−3.89 eV), consistent with the weaker electron‐deficient character of the modified terminal groups and intramolecular charge transfer effect. Density functional theory (DFT) calculations further corroborate these observations, showing that dTPE forms a favorable cascade energy alignment with D18 and L8BO‐C4 (Figures 1d; S1 and S2), which can facilitate charge separation and transport in devices. Thermogravimetric analysis (TGA) reveals that dTPE exhibits a higher thermal decomposition temperature (330.1°C) than L8BO‐C4 (322.4°C) (Figure S31a). Furthermore, the glass transition temperatures (T _g) were obtained from the temperature‐dependent UV–vis absorption spectra. As illustrated in Figure S31b, the T _g of dTPE, L8BO‐C4, and L8BO‐C4:dTPE films are calculated to 125.3°C, 121.5°C, and 122.3°C, respectively. These results demonstrate that dTPE exhibits superior thermal properties compared to L8BO‐C4.

Fluorescence spectroscopy was used to probe the luminescence characteristics of dTPE in comparison with conventional Y‐series acceptors. In dilute tetrahydrofuran solution, both dTPE and L8BO‐C4 exhibited weak emission in the 500–650 nm region (Figures 1e; S4). Upon gradual addition of water as a poor solvent to induce aggregation, a striking divergence emerged. dTPE developed a pronounced emission peak centered at ∼530 nm, and its photoluminescence intensity increased tenfold at a THF/water ratio of 1:3. By contrast, the emission of L8BO‐C4 remained essentially unchanged under identical conditions (Figure 1f), confirming the presence of AIE behavior in dTPE. Additional excitation at 530 nm further underscored these differences (Figure S5). Upon addition of water, L8BO‐C4 displayed severe ACQ, with its emission intensity dropping to only 5.3% of the original. In contrast, dTPE retained 72.9% of its fluorescence intensity, which remained stable with higher water fractions. This resilience can be attributed to the sterically hindered TPE unit, which suppresses detrimental π–π stacking and alleviates ACQ. The impact of this design is directly reflected in the solid‐state PLQYs. As summarized in Figures 1g and S6 and S7, the film PLQYs of L8BO‐C4, dTPE, and dSpiro were 1.33 ± 0.13%, 4.69 ± 0.19%, and 2.43 ± 0.12%, respectively. The combination of AIE activation and ACQ suppression thus enables dTPE to achieve a more than threefold enhancement in solid‐state luminescence relative to L8BO‐C4. Such an improvement is expected to directly mitigate non‐radiative recombination losses in organic solar cells, as the ΔE _nr is quantitatively linked to the EQE_EL through the relation ΔE _nr = −kT ln(EQE_EL). Since EQE_EL in turn depends strongly on the PLQY of the active layer, the solid‐state luminescence of dTPE provides a direct pathway toward reducing ΔE _nr at the molecular level.

Organic solar cell devices were fabricated in a conventional architecture of ITO/2PACz/D18:A1:(A2)/PDINN/Ag, with details provided in the supporting information. The current density–voltage (J–V) characteristics and photovoltaic parameters are shown in Figures 2a, S9, and summarized in Table 1. Among the binary devices, D18:dTPE yielded a moderate PCE of 8.21%, with a short‐circuit current density (J _SC) of 13.69 mA cm^{− 2}, a fill factor (FF) of 56.4%, and an outstanding V _OC of 1.069 V. By contrast, D18:dSpiro delivered a much lower PCE of 2.63% with inferior J _SC (6.17 mA cm^{− 2}), FF (42.1%), and V _OC (1.005 V), which can be attributed to its elevated HOMO level and unfavorable exciton dissociation. These results indicate that while the sterically hindered TPE group compromises charge transport in neat blends, it simultaneously endows dTPE with an exceptionally high V _OC, underscoring its huge potential as a guest acceptor in high‐performance systems. Therefore, dTPE was introduced into the benchmark D18:L8BO‐C4 system. As the dTPE fraction increased, the ternary devices displayed a systematic rise in V _OC from 0.886 V to as high as 1.069 V (Figure 1d; Table S1), while the FF remained stable at low guest contents but decreased at higher loadings. Consequently, the PCE exhibited a volcano‐like dependence on composition. The optimized ternary blend (1:1:0.2) achieved a champion PCE of 20.51%, surpassing the 19.37% of the D18:L8BO‐C4 binary device. External quantum efficiency (EQE) spectra (Figure 2b) confirmed the origin of the performance enhancement. The ternary device exhibited a slightly blue‐shifted spectrum and a modest increase in the short‐wavelength region, resulting in a J _SC of 27.51 mA cm^{− 2}, consistent with the integrated photocurrent (discrepancy < 5%), validating the accuracy of the J–V measurements. Reproducibility was verified through statistical analysis (Figure 2c). Notably, a blend ratio of 1:0.9:0.3 also delivered a high PCE of 20.29% while maintaining an elevated V _OC of 0.944 V, illustrating the robustness of the design. Importantly, the record‐high V _OC of 0.944 V for the optimized ternary device sets a new benchmark for organic solar cells with PCEs exceeding 20% (Figure 2e; Table S7).

FIGURE 2.

(a) Characteristic J–V curves of the optimized binary and ternary devices under simulated AM 1.5G irradiation (100 mW cm⁻²). (b) The corresponding EQE spectra and integrated J _SC values of binary and ternary devices. (c) Efficiency‐distributed box plots of the binary and ternary devices derived from 12 independent devices. (d) The dependence curves of V _OC, FF, and PCE of the ternary devices on the dTPE content. (e) Efficiency‐distributed box plots of the binary and ternary devices based on D18:acceptor active layer. (e) Plots of the PCE versus V _OC of the devices reported in 2025. (f) The J–V curves of binary and ternary devices in dark conditions. (g) Photocurrent density versus effective bias characteristics, (h) J _SC versus P, and (i) V _OC versus P characteristics of the binary and ternary devices. In this figure, the D18:L8BO‐C4:dTPE ternary device was prepared with a blend ratio of 1:1:0.2.

TABLE 1.

Summary of the detailed photovoltaic parameters of the binary and ternary OSCs.

Active layer	V OC [V]	JSC [mA cm−2]	J cal [mA cm−2]	FF [%]	PCE(PCE a ) [%]
D18: L8BO‐C4	0.886 (0.885 ± 0.002)	27.33 (27.23 ± 0.14)	26.34	80.2 (79.8 ± 0.6)	19.37 (19.17 ± 0.20)
D18: dTPE	1.069 (1.067 ± 0.003)	13.69 (13.49 ± 0.21)	13.38	56.4 (56.0 ± 0.5)	8.21 (7.98 ± 0.23)
D18: dSpiro	1.005 (1.003 ± 0.002)	6.17 (6.03 ± 0.15)	5.83	42.1 (41.9 ± 0.4)	2.63 (2.51 ± 0.12)
D18: L8BO‐C4: dTPE (1:1:0.2)	0.929 (0.926 ± 0.003)	27.51 (27.34 ± 0.19)	26.41	80.2 (79.6 ± 0.7)	20.51 (20.34 ± 0.17)

^a Average values obtained from 12 devices.

To gain further insights into exciton dissociation and charge collection, the photocurrent density (J _ph) as a function of the effective voltage (V _eff) was analyzed. As shown in Figure 2g, the charge collection efficiencies (η) of the D18:L8BO‐C4, D18:dTPE, and D18:L8BO‐C4:dTPE devices were determined to be 99.0%, 82.5%, and 99.3%, respectively. The relatively low η of the D18:dTPE device reflects its limited charge transport arising from the less favorable active layer morphology. In contrast, the ternary device achieves a slightly higher η than the binary counterpart, demonstrating that the incorporation of dTPE facilitates exciton dissociation and charge collection, in good agreement with its improved J _SC. To investigate recombination behavior, dark J–V and light‐intensity–dependent measurements were carried out. In OSCs, dark current originates primarily from carrier recombination within the active layer or at the interfaces [41, 42, 43]. As shown in Figure 2f, the dark current densities (J _d) at −2 V bias for the D18:L8BO‐C4, D18:dTPE, and D18:L8BO‐C4:dTPE devices were 2.67 × 10^{− 4}, 3.26 × 10^{− 6}, and 1.89 × 10^{− 5} A cm^{− 2}, respectively. The ternary device therefore exhibits a J _d one order of magnitude lower than that of the binary device, highlighting the suppression of non‐radiative recombination and the consequent reduction in energy loss. The recombination kinetics were further analyzed by monitoring the dependence of J _SC and V _OC on light intensity (P) (Figure S12). The J _SC–P relationship follows J _SC ∝ P ^α, where α characterizes the degree of bimolecular recombination [44, 45, 46]. As shown in Figure 2h, the extracted α values for D18:L8BO‐C4, D18:dTPE, and D18:L8BO‐C4:dTPE devices are 0.962, 0.933, and 0.968, respectively. The α value of the ternary device is closer to unity than that of the binary device, indicating its reduced bimolecular recombination. For V _OC–P analysis, the slope of V _OC versus ln(P) indicates the extent of trap‐assisted recombination [47, 48, 49]. As presented in Figure 2i, the slopes of the D18:L8BO‐C4, D18:dTPE, and D18:L8BO‐C4:dTPE devices are 1.14, 1.23, and 1.12 kT/q, respectively, confirming the suppression of trap‐assisted recombination in the ternary system. Charge transport characteristics were evaluated using SCLC method [50, 51]. As presented in Figures S10 and S11, the µ _e and µ _h of the D18:dTPE device are 3.14 × 10^{− 4} and 7.64 × 10^{− 4} cm² V^{− 1} s^{− 1}, respectively. For the D18:L8BO‐C4 binary film, µ _e and µ _h are 4.66 × 10^{− 4} and 6.49 × 10^{− 4} cm² V^{− 1} s^{− 1}, with a µ _h/µ _e ratio of 1.39. Incorporating dTPE increases both carrier mobilities to 5.67 × 10^{− 4} (µ _e) and 7.55 × 10^{− 4} cm² V^{− 1} s^{− 1} (µ _h), while reducing the µ _h/µ _e ratio to 1.30. The detailed parameters are summarized in Table S2. These results suggest a more balanced and efficient charge transport in the ternary blend. Taken together, the higher exciton dissociation efficiency, balanced carrier mobilities, and suppressed recombination losses in the D18:L8BO‐C4:dTPE device account for its superior photovoltaic performance compared to the binary control.

To elucidate the origin of the high V _OC in TPE‐based devices, we conducted a comprehensive energy loss (E _loss) analysis (Figure S13 and Table 2). As shown in Figure 3a, the EQE_EL values for D18:L8BO‐C4, D18:dTPE, D18:dSpiro, and D18:L8BO‐C4:dTPE are 2.29 × 10^{− 4}, 5.68 × 10^{− 3}, 3.53 × 10^{− 4}, and 5.39 × 10^{− 4}, respectively. Compared to L8BO‐C4, dSpiro benefits from its bulky end group, which weakens molecular packing and suppresses ACQ, thereby slightly enhancing its EQE_EL. Remarkably, despite sharing a sterically bulky end group motif with dSpiro, dTPE delivers an EQE_EL as high as 0.57%, more than an order of magnitude higher than both dSpiro and L8BO‐C4. This EQE_EL value is among the best reported for state‐of‐the‐art OSC acceptors (Figure S14), and originates from the intrinsic AIE behavior of dTPE, which promotes strong solid‐state luminescence. As a result, the D18:dTPE binary device achieves an exceptionally low E _loss of 0.452 eV, significantly lower than that of D18:L8BO‐C4 (0.543 eV) and D18:dSpiro (0.517 eV). Incorporation of 20 wt.% dTPE into the D18:L8BO‐C4 host further reduces E _loss from 0.543 to 0.516 eV in the ternary device. Figure 3b highlights that suppression of ΔE _nr is the dominant factor underlying this reduced E _loss. It is widely recognized that minimizing ΔE _nr is one of the most critical yet challenging tasks in pushing OSC performance toward its theoretical limit, and only a few molecular design strategies have proven effective [8, 54, 55, 56, 57]. Here, dTPE achieves a ΔE _nr as low as 0.130 eV, representing the lowest non‐radiative loss reported to date in OSCs (Figure 3c; Table S6). The ternary device also benefits, with ΔE _nr reduced to 0.190 eV compared to 0.219 eV for the binary control. These results provide compelling evidence that the AIE molecular design is an effective pathway to enhance luminescence in non‐fullerene acceptors and simultaneously suppress non‐radiative recombination losses in OSCs. We further evaluated energetic disorder through the Urbach energy (E _U), extracted from Fourier‐transform photocurrent spectroscopy (FTPS). As shown in Figure 3d, the optimized ternary blend exhibits a E _U of 23.22 meV, significantly lower than that of D18:L8BO‐C4 (25.06 meV), D18:dTPE (26.05 meV), and D18:dSpiro (28.07 meV). Electrochemical impedance spectroscopy was also performed to investigate the shunt resistance (R _sh). As shown in Figure S34, the R _sh value of D18:L8BO‐C4:dTPE‐based device (17.3 kΩ) is higher than that of D18:L8BO‐C4‐based device (15.1 kΩ). The reduced E _U and enhanced R _sh indicates that the incorporation of dTPE promotes more ordered π–π stacking in the blend film, further contributing to suppressed energetic disorder and improved V _OC.

TABLE 2.

Detailed energy loss analysis of the optimized ternary and binary devices.

Active layer	E g a [eV]	qV OC [eV]	Eloss [eV]	qVOC SQ b [eV]	ΔE 1 [eV]	ΔE 2 [eV]	ΔE nr c [eV]	EQE EL	EU d [meV]
D18: L8‐BOC4	1.425	0.882	0.543	1.164	0.261	0.066	0.219	2.29 × 10−4	25.06
D18:dTPE	1.516	1.064	0.452	1.251	0.265	0.065	0.130	5.68 × 10−3	26.05
D18:dSpiro	1.522	1.005	0.517	1.259	0.263	0.071	0.194	3.53 × 10−4	28.07
D18:L8‐BOC4:dTPE	1.445	0.929	0.516	1.181	0.264	0.066	0.190	5.39 × 10−4	23.22

^a E _g values are deduced from the intercrossing of the normalized absorption and emission spectra.

^b qV _OC ^SQ is the maximum V _OC by the SQ limit.

^c ΔE_nr is determined from the EQE_EL.

^d E _U is commonly‐used to evaluate the energetic disorder degree in organic semiconductor device, which can be calculated from the EQE band tails [52, 53].

FIGURE 3.

(a) EQE_EL as a function of injection current and the (b) statistical diagram of energy loss for the binary and ternary devices. (c) Plots of the ΔE ₃ versus E _loss of the OSCs reported in the literatures. (d) Extraction of Urbach energy from ln(EQE) in the long‐wavelength edge.

Atomic force microscopy (AFM) was employed to probe the surface morphology of the active layers. As shown in Figure 4a–d, all four blend films exhibit well‐defined fibrillar features. The binary D18:L8BO‐C4 film displays a root‐mean‐square roughness (R _q) of 1.88 nm, while both the D18:dTPE (R _q = 1.23 nm) and D18:dSpiro (R _q = 1.26 nm) films exhibit lower roughness values, which can be attributed to their relatively weaker crystallinity. Interestingly, incorporation of 20 wt.% dTPE into the D18:L8BO‐C4 host results in a ternary film with a reduced roughness of 1.56 nm, accompanied by enlarged fiber dimensions and a more favorable fibrous interpenetrating network. This morphology is consistent with the enhanced charge transport observed in the ternary devices, supporting more efficient exciton dissociation and carrier collection. To gain deeper insight into molecular packing, grazing‐incidence wide‐angle X‐ray scattering (GIWAXS) measurements were carried out on both neat and blend films (Figures 4e–i; S15). The three acceptor neat films all adopt a preferential face‐on orientation, with distinct (010) diffraction peaks in the out‐of‐plane (OOP) direction (Table S4). The q _z positions of these peaks correspond to π–π stacking distances of 3.56 Å (L8BO‐C4), 3.86 Å (dTPE), and 3.89 Å (dSpiro), confirming that the bulky end groups of dTPE and dSpiro relax their molecular packing. Coherence length (CL) analysis further indicates weaker crystallinity in dTPE (31 Å) and dSpiro (17 Å) compared with L8BO‐C4 (36 Å). The blend films exhibit similar face‐on orientation, but with notable differences in crystallinity (Table S5). The D18:L8BO‐C4 film shows an out‐of‐plane (010) coherence length of 26 Å, which decreases to 23 and 20 Å in the D18:dTPE and D18:dSpiro films, respectively, consistent with their less crystalline nature. Unexpectedly, however, the D18:L8BO‐C4:dTPE ternary blend shows π–π stacking distance reduced from 3.62 to 3.58 Å and the CL increased from 26 to 29 Å. The incorporation of dTPE promotes a more ordered stacking at D18/L8BO‐C4 interface, resulting in shorter π‐π stacking distances. This observation is intriguing, as dTPE itself displays weaker crystallinity and looser packing in its neat film.

FIGURE 4.

The AFM height images and phase images of (a) D18: L8BO‐C4, (b) D18: dTPE, (c) D18: dSpiro, and (d) D18: L8BO‐C4: dTPE blended films. The 2D GIWAXS profiles of (e) D18: L8BO‐C4, (f) D18: dTPE, (g) D18: dSpiro, (h) D18: L8BO‐C4: dTPE blended films, and i) the corresponding 1D line‐cut curves along OOP and IP directions of the binary and ternary blended films.

To gain further mechanistic understanding of why the ternary blend exhibits improved order despite the bulky TPE group as well as to elucidate the origin of restricted intramolecular rotation (RIR), molecular dynamics (MD) simulations were carried out on both binary (D:A1) and ternary (D:A1:A2) systems. Intermolecular packing was evaluated using the radial distribution function, g(r). As shown in Figures 5b and S18, the g(r) between the donor and acceptor in the ternary blend exhibits a markedly sharper first peak at 3.6 Å compared to the weaker peak at 3.8 Å in the binary blend. This result indicates that the incorporation of dTPE promotes more ordered packing and a shorter π–π stacking distance at the D:A interface, consistent with the GIWAXS analysis. We then examined the local environment of the TPE moieties of dTPE to understand the origin of the RIR effect. The g(r) analysis in Figure 5c reveals two dominant interactions including TPE self‐aggregation (TPE(A2):TPE(A2)) and stacking between TPE and the acceptor moiety of the donor (TPE(A2):A_moiety(D)). Representative snapshots of these configurations are displayed in Figure 5d. In both cases, adjacent molecular fragments intercalate between the phenyl rings of TPE, creating steric hindrance that restricts intramolecular torsional motion. This restriction is confirmed by a narrower torsion angle distribution (Figure 5e), with the standard deviation (σ_θ) decreasing from 1.7° in solution to 1.2° in the blend. Such restricted intramolecular rotation serves as the molecular basis for activating aggregation‐induced emission while simultaneously stabilizing intermolecular packing in the ternary system. These results demonstrate that dTPE, as a guest acceptor, can not only reduce the ΔE _nr of OSCs, but also enhance device performance by optimized blended morphology.

FIGURE 5.

MD simulations revealing the restricted intramolecular rotation of dTPE in ternary blend. (a) The donor moiety (D _moiety), acceptor moiety (A _moiety), and TPE moiety of dTPE (denoted as A2). (b) g(r) between the D _moiety of donor and the D _moiety of acceptor in binary (black lines) and ternary (red lines) blends. The D _moiety of acceptor in binary blend and ternary blend are denoted as D_m(A1) and D _m(A1+A2), respectively. (c) g(r) between the TPE moiety of A2 and the other moieties. (d) Representative snapshots of the TPE(A2):TPE(A2) and TPE(A2):A _m(D) stacking configurations extracted from the MD simulated ternary blend. (e) Gaussian fitted distributions of the average intramolecular torsion angles (θ) of the TPE moieties in solution, TPE(A2):TPE(A2) stacking, and TPE(A2):A _m(D) stacking calculated from 500 frames of the MD trajectory. σ_θ represents the torsion angle disorder.

3. Conclusion

In summary, we have demonstrated that aggregation‐induced emission molecular design offers a powerful approach to suppressing non‐radiative recombination losses in OSCs. By introducing TPE units into the Y‐series acceptor, dTPE exhibits AIE behavior that suppresses ACQ and enhances film‐state PLQY. This molecular‐level advancement directly translates into a high EQE_EL and a remarkable non‐radiative recombination loss as low as 0.130 eV in binary devices, as well as a record V _OC of 0.93 V with a ΔE _nr of 0.190 eV and a PCE of 20.5% in ternary blends. Beyond these record metrics, this study highlights a generalizable strategy that bridges the molecular photophysics of AIE with photovoltaic performance optimization. We believe this work not only enriches the design toolbox for high‐efficiency OSCs but also points toward a future where molecular design of luminescent acceptors with enhanced PLQY becomes a central strategy for suppressing non‐radiative losses.

Conflicts of Interest

The authors declare no conflicts of interest.

Supporting information

Supporting File 1: adma72166‐sup‐0001‐Data.zip.

Supporting File 2: adma72166‐sup‐0002‐SuppMat.docx.

Data Availability Statement

The data that support the findings of this study are available in the supplementary material of this article.