High‐Efficiency Organic Solar Cells Enabled by Siloxane‐Functionalized Pyrazine Terpolymers: Synergizing Performance, Morphology Control, and Non‐Halogenated Solvent Processability

ABSTRACT

Balancing high performance, morphological controllability, and compatibility with non‐halogenated solvent processing remains a critical bottleneck for scalable and sustainable organic solar cells (OSCs). Herein, we address this challenge via rational terpolymer design: integrating a siloxane‐functionalized electron‐deficient pyrazine unit (DTCPz‐SiO) into the benchmark D18 backbone, with the optimized terpolymer DN1 containing 5 mol% DTCPz‐SiO. DTCPz‐SiO imparts two key synergies: (i) enhanced conformational rigidity and intramolecular noncovalent interactions (N···S, N···H), which improve backbone planarity, strengthen π–π stacking, and accelerate crystallization; (ii) synergistic regulation of donor‐acceptor miscibility and compatibility with non‐halogenated solvents. These effects collectively enable a well‐optimized bulk‐heterojunction morphology with enhanced molecular ordering and charge dynamics. Consequently, DN1‐based binary devices deliver a significantly improved power conversion efficiency (PCE) of 20.1% compared to 18.7% for the parent polymer, together with a broadened processing window. Notably, high efficiencies of ∼19.5% are retained under common non‐halogenated processing conditions. Furthermore, DN1‐based ternary OSCs enhance PCE to outstanding values of 20.9% and 20.0% under chlorinated and non‐halogenated processing conditions, respectively, among the highest efficiencies reported for single‐junction OSCs. Overall, this work establishes siloxane‐functionalized terpolymers as an effective molecular design strategy for regulating multi‐scale morphology and processing tolerance, providing new insights for the development of scalable OSC systems.

Incorporation of a siloxane‐functionalized, conformationally rigid pyrazine unit (DTCPz–SiO) into the D18 backbone affords a multifunctional terpolymer (DN1) that synergistically enhances backbone planarity, optimizes morphology, and enables non‐halogenated solvent processing. The resulting organic solar cells achieve > 20% efficiency under both halogenated and non‐halogenated solvents, bridging high performance with sustainable fabrication.

1. Introduction

Organic photovoltaics (OPVs) have advanced rapidly, driven by the development of high‐performance donor‐ (D‐)/acceptor‐ (A‐) materials and a deeper understanding of the structure‐function relationships governing bulk heterojunction (BHJ) film morphology [1, 2, 3, 4, 5, 6, 7]. Recent breakthroughs in molecular design and device engineering have pushed organic solar cells (OSCs) power conversion efficiencies (PCEs) beyond 20% [8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18]. Central to these gains are high‐efficiency D‐polymers that pair with state‐of‐the‐art non‐fullerene acceptors (NFAs) to facilitate effective charge generation, transport, and collection. Among them, a prominent example is the polymer D18 [19], which has attracted widespread attention for its impressive performance due to its high hole mobility, well‐aligned energy levels, and excellent morphological compatibility with NFAs [20]. However, D18 suffers from a critical limitation: its pronounced aggregation tendency and narrow processing window hinder device reproducibility and large‐scale manufacturing, especially under environmentally friendly, non‐halogenated solvent conditions [9, 21, 22, 23]. This challenge exposes a core bottleneck in OPV development: the need for molecular design strategies that simultaneously preserve high photovoltaic performance, enhance morphological controllability, and enable compatibility with non‐halogenated solvent processing.

Random terpolymerization has emerged as a versatile strategy to tune polymer properties—including solubility, crystallinity, energy levels, and intermolecular packing—while retaining the parent backbone's advantages [22, 24, 25, 26, 27, 28, 29, 30, 31]. Incorporating electron‐deficient, conformationally rigid third comonomers can improve backbone planarity, strengthen π–π interactions, and broaden processing windows (even for non‐halogenated solvents). [24, 28, 29] However, existing terpolymer strategies often fall short: they rarely address the synergistic optimization of efficiency, morphology control, and non‐halogenated solvent processability. Critically, OSC performance depends not just on molecular structure, but on BHJ film formation kinetics‐subtle changes in polymer architecture (e.g., backbone stiffness, pre‐aggregation behavior) can disrupt film formation: premature aggregation causes excessive phase separation, while delayed crystallization leads to disordered morphologies that impede charge transport [32, 33, 34, 35]. A precise, integrated approach to tuning both molecular structure and film formation kinetics remains lacking.

To address this gap, we report a rational terpolymer design strategy: region‐random incorporation of a siloxane‐functionalized, conformationally rigid, electron‐deficient unit (DTCPz‐SiO) into the D18 backbone, yielding DN1 (5 mol%) and DN2 (10 mol%). This design simultaneously optimizes photoelectronic properties, aggregation behavior, and processing compatibility. The optimized terpolymer, DN1, delivers superior performance, achieving a PCE of 20.1% under chloroform processing and retains ≈19.5% efficiency with non‐halogenated solvents. The DTCPz‐SiO unit acts as a multi‐functional modulator: (1) its conformational rigidity and intramolecular noncovalent interactions (N···S, N···H) enhance backbone planarity and strengthen π–π stacking while suppressing over aggregation [36]; (2) it improves compatibility with non‐halogenated solvents and fine‐tunes donor‐acceptor miscibility; (3) it deepens the highest occupied molecular orbital (HOMO) energy level, boosting open‐circuit voltage (V _OC); (4) incorporation of DTCPz‐SiO accelerates crystallization while mitigating the terpolymer‐induced reduction in lamellar ordering, thus maintaining favorable molecular packing for efficient charge transport. These synergistic effects yield optimized, fiber‐like phase‐separation morphology in the BHJ, enabling efficient interfacial charge transfer, balanced hole/electron mobilities, and suppressed nonradiative recombination. Remarkably, ternary OSCs based on DN1:L8‐BO:AITC further enhance PCEs of 20.9% and 20.0% under halogenated and non‐halogenated conditions, respectively. This minimal PCE loss (<5%) outperforms most non‐halogenated solvent‐processable OSCs and aligns with industrial roll‐to‐roll manufacturing needs. Overall, integrating siloxane‐tethered, conformationally rigid building blocks into donor polymers resolves the long‐standing trade‐off between efficiency, morphology control, and non‐halogenated solvent processability, offering a generalizable framework for molecular engineering and sustainability of OPV design.

2. Results and Discussion

The molecular structures of D18 and the two novel terpolymers, DN1 and DN2, are shown in Figure 1a. These terpolymers were synthesized via copolymerization of three monomeric units: an electron‐donating group based on benzodithiophene (BDT), and two electron‐accepting components, DTBT [19] and the newly developed DTCPz‐SiO unit. As shown in Figure 1b, the brominated monomer DTCPz‐SiO‐Br was synthesized through a three‐step procedure starting from compound 1, which was previously reported by our group [36]. The synthetic sequence involved: (i) a transesterification reaction to yield a vinyl‐terminated combinatorial side chain; (ii) a siloxanation reaction to introduce a siloxane‐terminated side chain; and (iii) bromination with N‐bromosuccinimide (NBS) to afford the final dibrominated DTCPz‐SiO‐Br monomer. The chemical structures of the intermediates were confirmed via ¹H/¹³C NMR (Figures S32–S35). Subsequently, polymers DN1 and DN2 were successfully synthesized by Stille coupling polymerization using Pd₂(dba)₃ and P(o‐tol)₃ as the catalytic system, with the feed ratio of DTBT to DTCPz‐SiO carefully controlled (Scheme S1). Detailed synthetic procedures and comprehensive structural characterization are provided in the Supporting Information. D18 and two terpolymers exhibited comparable number‐average molecular weights (M _n) ranging from 42.2 kDa to 47.6 kDa, with polydispersity indices (PDI) close to 2.0, as determined by gel permeation chromatography (GPC) (Figures S36–S38). Furthermore, thermogravimetric analysis (TGA), performed under a nitrogen atmosphere, revealed thermal decomposition temperatures (T _d, defined as 5% weight loss) of approximately 431°C for D18, 428°C for DN1, and 422°C for DN2 (Figure S1), indicating sufficient thermal stability for application in OSC fabrication.

FIGURE 1.

(a) Molecular structures of D18, DN1, and DN2. (b) The synthetic routes for the monomer DTCPz‐SiO‐Br. (c) rPES scan of the BDT‐DTBT backbone (representing D18), BDT‐DTCPz backbone (representing terpolymer), showing the relatively rotational profiles of dihedral angles r₁, r₂, r₃, r_4, and r₅ as indicated in (d,e). (d,e) Optimized geometry of these backbones, highlighting the dihedral angles r₁, r₂, r₃, r₄, and r₅ at the minimum energy conformations. (f) Energy level diagram of D18, DN1, DN2, and L8‐BO neat films, and (g) their corresponding normalized UV–vis absorption spectra. (h) The various I _0‐0/I _0‐1 values of the polymer donors at different temperatures. (i) J–V curves with a schematic diagram of the device architecture and (j) EQE spectra and integrated J _SC of OSCs based on D18, DN1, and DN2 as donors.

To investigate the molecular geometries of the terpolymers in comparison to the parent polymer D18, electronic structure calculations were carried out using standard density functional theory (DFT), at B3LYP/6‐311G(d,p) level of theory, with GD3BJ empirical dispersion corrections, as detailed in Supporting Information. To reduce computational cost, all long alkyl side chains were replaced with methyl groups in this initial set of calculations. In the simulated models, the thiophene π‐bridge between BDT and dithieno[3′,2′:3,4;2″,3″:5,6]benzo[1,2‐c][1,2,5]thiadiazole (BT) or carboxylate‐substituted pyrazine (CPz) units may undergo rotational motion around single C─C bonds, leading to multiple possible conformations. To explore the conformational landscape, a relaxed potential energy surface (rPES) scan was carried out by systematically rotating the dihedral angles from r₁, to r₅, as shown in Figure 1c. Interestingly, the BDT‐DTBT‐based system exhibits energy minima that are closely spaced (≤ 0.75 kcal mol⁻¹) and separated by barriers lower than 4 kcal mol⁻¹. This indicates a larger number of accessible conformers, with three distinct local minima identified for each dihedral rotation (r₁ and r₂). In contrast, the BDT‐DTCPz‐based system also shows minima separated by small energy differences (≤ 1.25 kcal mol⁻¹), but these are divided by comparatively higher energy barriers, particularly evident for r₄ and r₅ (achieving almost 8 kcal mol⁻¹ for r₄), reflecting a greater degree of conformational rigidity. These results suggest that the BT‐based polymer backbone exhibits greater conformational diversity, whereas the CPz‐based backbone is more conformationally constrained. The incorporation of the DTCPz‐SiO unit thus enhances the structural stability of the polymer backbone and reduces the likelihood of thermally induced torsional fluctuations, favoring more planar geometry.

Furthermore, analysis of the optimized lowest‐energy geometries (Figure 1d,e) reveals that the DTCPz‐containing backbone adopts a significantly flatter molecular structure, with an average dihedral angle of approximately 10.5°, compared to 21.5° observed in D18. As reported in our previous studies [29, 36], the incorporation of DTCPz introduces noncovalent intramolecular interactions (such as N···S and N···H), which synergistically enhance backbone planarity and promote ordered stacking. Consequently, this pronounced difference in molecular planarity is expected to significantly influence the aggregation behavior and intermolecular packing of the materials in the solid state, which will be discussed in detail in the following sections.

In the calculations involving frontier molecular orbitals (FMOs) and natural transition orbitals (NTOs), the optimally tuned DFT approach [37, 38] was employed in order to account for long‐range electron‐electron exchange interactions under the electrostatic film environment conditions. Consult Supporting Information for more details. This time, the siloxane‐functionalized side chain was retained due to its potential influence on the electronic structure (Figure S2). The calculated energy levels indicate that the terpolymers exhibit downshifted energy levels compared to D18. This shift is attributed to the strong electron‑withdrawing character of the 1,4‑diazine ring, arising from its C═N bonds, combined with the additional electron‑withdrawing effect of the ester group. These computational results align well with experimental observations from cyclic voltammetry measurements (Figure S3), which indicate E_{LUMO (lowest unoccupied molecular orbital)}/E_HOMO values of −3.58/−5.38, −3.56/−5.40, and −3.57/−5.43 eV for D18, DN1, and DN2, respectively (Figure 1f). The observed deeper HOMO energy level in the target polymer is advantageous for achieving higher V _OC in photovoltaic devices.

The optical properties of the D‐polymers were investigated in dilute chloroform solution and in thin films using UV–vis absorption spectroscopy (Figure 1g; Figure S4 and Table S1). In solution, D18, DN1, and DN2 displayed similar absorption profiles with clearly resolved vibronic structures (Figure S4). Among them, D18 exhibited the strongest 0‐0 peak in the 450–600 nm region, indicating a greater degree of molecular aggregation compared to the terpolymer analogues. In thin film, D18 showed enhanced 0–1 vibronic shoulder and suppressed 0‐0 transitions (Figure 1g), consistent with increased interchain coupling and a tendency toward H‐type aggregation [39, 40]. Remarkably, as the content of the DTCPz‐SiO unit increased, the 0‐0 peak intensity in DN1 and DN2 re‐emerged and became more prominent, accompanied by a narrowing of the optical bandgap. These spectral changes indicate more pronounced J‐type aggregation and enhanced π–π interactions in the solid state [41, 42], arising from improved intermolecular ordering introduced by the planar DTCPz‐SiO unit. In particular, DN1 exhibited the highest absorption coefficient, benefiting from an optimal balance of backbone planarity and moderate aggregation that maximizes π‐conjugation and photon‐harvesting (Figure S5). In contrast, DN2 shows a lower absorption coefficient, likely due to excessive aggregation and sequence‐induced energetic disorder that reduces the fraction of well‐ordered conjugated domains.

To further probe differences in aggregation behaviour, temperature‐dependent UV–vis absorption spectra were recorded in chlorobenzene (Figure S6). As shown in Figure 1h, the intensity ratio of the 0‐0 to 0‐1 vibronic peaks (I _0‐0/I _0‐1) decreased with increasing temperature for all samples, with D18 showing the most rapid decline. This trend suggests that D18 aggregates more strongly at lower temperatures but is also more prone to thermal dissociation. Below 40°C, D18 had the highest I _0‐0/I _0‐1 ratio. However, between 40°C and 50°C, its aggregation became comparable to that of DN1, and at 100°C, it was lower, confirming the weaker thermal stability of its aggregates, indicating that DTCPz‐SiO incorporation thermally controls the torsional fluctuations by a higher degree of conformational rigidity. Based on these results, a fabrication temperature of 45°C was chosen to ensure similar pre‐aggregate degrees for D18 and DN1 during device processing. Under these conditions, OSCs incorporating the terpolymer Ds and L8‐BO as the A exhibited markedly improved performance. Devices based on terpolymers, particularly those with 5 mol% of the modified unit, showed simultaneous increases in short‐circuit current density (J _SC), V _OC, and fill factor (FF), resulting in an overall enhancement in PCE of approximately 7.5% relative to the D18‐based reference (Table 1).

TABLE 1.

Photovoltaic parameters of OSCs fabricated from chloroform under AM 1.5 G illumination (100 mW cm⁻²).

System	V OC a (V)	J SC a (mA cm−2)	J cal b (mA cm−2)	FF a (%)	PCE a (%)
D18:L8‐BO	0.901 (0.902 ± 0.002)	26.5 (26.4 ± 0.2)	25.4	78.3 (78.1 ± 0.3)	18.7 (18.6 ± 0.1)
DN1:L8‐BO	0.918 (0.916 ± 0.002)	27.1 (27.2 ± 0.1)	26.1	80.6 (80.3 ± 0.2)	20.1 (20.0 ± 0.1)
DN2:L8‐BO	0.926 (0.927 ± 0.001)	26.0 (25.8 ± 0.1)	24.9	77.7 (77.5 ± 0.2)	18.7 (18.5 ± 0.1)
DN1:L8‐BO:AITC	0.931 (0.931 ± 0.002)	27.7 (27.5 ± 0.1)	26.6	80.9 (81.0 ± 0.2)	20.9 (20.7 ± 0.2)
DN1:L8‐BO:AITC c	0.911 (0.909 ± 0.002)	27.6 (27.5 ± 0.1)	26.3	79.7 (79.6 ± 0.2)	20.0 (19.9 ± 0.1)

^a The values in parentheses were averaged from ten independent cells.

^b Calculated from the corresponding EQE spectrum.

^cOSCs fabricated from o‐XY:CS₂ = 1:1 (v/v).

All devices were fabricated and characterized using a conventional sandwich structure comprising ITO/PEDOT:PSS/Donor:L8‐BO/PNDIT‐F3N/Ag. The current density–voltage (J–V) characteristics and key parameters of the champion devices are presented in Figure 1i and summarized in Table 1. The D18:L8‐BO control device delivers a PCE of 18.7% with a V _OC of 0.901 V, J _SC of 26.5 mA cm⁻², and an FF of 78.3%. With the introduction of DTCPz‐SiO units into the polymer backbone, a systematic modulation of device performance was observed. The V _OC increased progressively with higher DTCPz‐SiO content, reaching 0.918 V for DN1 and 0.926 V for DN2, a trend attributable to the deepened HOMO energy levels of the terpolymers, as confirmed by cyclic voltammetry. In contrast, the J _SC and FF exhibited a non‐monotonic trend, both peaking in the DN1:L8‐BO device, which achieved a J _SC of 27.1 mA cm⁻² and an FF of 80.6%, culminating in a maximum PCE of 20.1% (certified as 19.83%, as shown in Figure S7). The J _SC values derived from external quantum efficiency (EQE) integration closely matched the J–V measurements, with deviations within ∼4%, supporting the reliability of the device data (Figure 1j). Additionally, we have further synthesized DN1 in two additional independent batches. The corresponding GPC data are presented in Figure S39, and the device performance data are summarized in Table S2. These results confirm the excellent material reproducibility and reliable device performance of DN1. However, further increasing the DTCPz‐SiO content to form DN2 led to diminished performance, with the PCE declining to 18.7%, comparable to that of the D18‐based device. This decrease is likely due to suboptimal active layer morphology or impaired charge transport, arising from excessive incorporation of the DTCPz‐SiO unit, an effect that will be examined in detail in the following sections.

To elucidate the origin of the enhanced J _SC and FF, the morphology of the active layer was investigated, as it plays a critical role in determining device performance. As previously discussed, DN1 and D18 were processed at comparable aggregation states, yet they exhibited markedly different film formation behaviors during the drying process. Figure 2a–c presents the time‐evolution contour plots of the in situ UV–vis absorption spectra for the three D‐polymers during film formation. Typically, the film formation process can be divided into three distinct stages: solvent evaporation (Stage I), nucleation and crystal growth (Stage II), and film drying (Stage III) [42, 43, 44]. During Stage I, minimal spectral shifts occur due to the rapid evaporation of the solvent. In Stage II, molecular self‐assembly initiates, followed by rapid nucleation and crystal growth once the critical supersaturation is reached. With increasing incorporation of the third component, DTCPz‐SiO, the terpolymers exhibited progressively earlier onset and shorter durations of the aggregation process (Stage II) (Figure 2d). Quantitatively, the onset of aggregation advanced from 1226 ms for D18 to 1179 ms for DN1, and further to 738 ms for DN2, accompanied by a corresponding reduction in aggregation duration from 284 ms to 251 ms and 137 ms, respectively (Table S3), indicating accelerated nucleation and crystallization kinetics. Consequently, the total film‐formation time was significantly reduced. This effect was modest in DN1 but became pronounced in DN2. According to prior studies [43, 45, 46, 47], earlier aggregation can typically result from one or more of the following factors: stronger intrinsic crystallinity, poorer solubility, or enhanced pre‐aggregation behavior in solution. In this case, temperature‐dependent absorption spectroscopy reveals that D18 exhibits the strongest pre‐aggregation behavior among the three polymers, effectively ruling out this factor (Figure S6). Moreover, the incorporation of random DTCPz‐SiO units into the terpolymer backbone improves solubility, excluding poor solubility as a likely cause. Therefore, the observed behavior is most plausibly attributed to the intrinsically higher crystallinity of the terpolymer materials.

FIGURE 2.

(a–c) Time‐resolved UV–vis absorption contour maps during film formation. (d) Temporal evolution of the donor's main absorption peak intensity. (e) Estimated d‐spacing and CCL values from the OOP (010) diffraction peaks of the polymer films. (f) Calculated face‐on to edge‐on orientation ratios from the (010) peak obtained from the 2D GIWAXS patterns. (g–i) 2D GIWAXS patterns of D18, DN1, and DN2 neat films.

Thus, the influence of DTCPz‐SiO incorporation on molecular ordering was carried out via grazing incidence wide‐angle X‐ray scattering (GIWAXS) measurements on neat polymer films. The resulting 2D GIWAXS patterns, along with the corresponding in‐plane (IP) and out‐of‐plane (OOP) line‐cut profiles, are shown in Figure 2g–i and Figure S8. All D‐polymers exhibited a predominant face‐on orientation, as evidenced by the pronounced (010) π–π stacking diffraction peaks in the OOP direction. Notably, a systematic trend was observed as the DTCPz‐SiO content increased: the π–π stacking distance progressively decreased from 3.84 Å for D18 to 3.77 Å for DN1 and 3.76 Å for DN2, while the corresponding crystal coherence length (CCL) [48] expanded markedly from 11.3 Å for D18 to 22.7 Å for DN1 and 22.8 Å for DN2 (Figure 2e; Table S4). These observations point to increasingly compact and ordered intermolecular packing, attributable to the backbone planarization effect imparted by the additional DTCPz‐SiO units. This improved packing order also accounts for the enhanced 0‐0 vibronic absorption peak intensity observed in the UV–vis spectra of the terpolymer thin films (Figure 1g), reflecting greater structural order in the solid state. Correlating with these structural insights, the increased crystallinity observed upon DTCPz‐SiO incorporation aligns well with the shortened aggregation times extracted from film formation kinetics. Notably, while DN1 achieves a favorable balance between crystallinity and kinetic ordering, DN2, with a higher DTCPz‐SiO content, exhibits an excessively rapid aggregation process, likely limiting the time available for optimal molecular self‐organization. This inference is supported by the reduced diffraction intensity observed in the 2D‐GIWAXS pattern of DN2 (Figure 2i), despite its narrower π–π stacking distance and longer CCL. Moreover, DTCPz‐SiO incorporation was also found to promote a higher degree of face‐on orientation, which is favorable for vertical charge transport in OSC devices. Quantitative analysis via azimuthal integration of the (010) diffraction signal (Figure S9) over the 45°–90° range revealed a continuous increase in the proportion of face‐on crystallites, from 54% in D18% to 62% in DN2 (Figure 2f). This enhancement is attributed to the synergistic effects of the planar DTCPz‐SiO moieties and the structural alignment facilitated by the siloxane‐functionalized side chains [49].

Building upon the insights obtained from neat films, we next examined how DTCPz‐SiO content influences the molecular organization within the active layer blend morphology. GIWAXS measurements on the BHJ films (Figure 3a; Figure S10) revealed distinct variations in lamellar ordering as a function of terpolymer composition. Specifically, the (100) diffraction peak at q _r = 0.32 Å⁻¹, corresponding to the lamellar stacking of D‐polymers in the IP direction, was most intense in the D18:L8‐BO blend (q _r = 0.41 Å⁻¹ assigned to L8‐BO; Figure S11), indicating the highest degree of long‐range order and the largest CCL (87.8 Å). However, this lamellar ordering became progressively attenuated upon partial substitution of the DTBT units with DTCPz‐SiO (Figure 3b). DN1:L8‐BO maintained a moderate level of lamellar crystallinity (CCL = 64.9 Å), whereas DN2:L8‐BO exhibited a pronounced reduction (CCL = 59.1 Å), reflecting disrupted molecular packing in the blend. To gain further insight into the origin of these morphological changes, we examined the film formation kinetics of the BHJ blends (Figure 3d–f). When processed with the high‐performance NFA L8‐BO, terpolymer donors showed accelerated aggregation behavior in their blends, with the aggregation duration decreasing from 131 ms (D18) to 97 ms (DN1) and 71 ms (DN2), relative to their neat counterparts. This effect was particularly prominent in DN2, where the significantly shortened aggregation duration is likely insufficient to support optimal nucleation and crystal growth, thereby contributing to the reduced lamellar crystallinity observed.

FIGURE 3.

(a) 1D line cut profiles along the OOP and IP directions extracted from 2D GIWAXS patterns of D18:L8‐BO, DN1:L8‐BO, and DN2:L8‐BO blend films. (b,c) Estimated d‐spacing and CCL values derived from the IP (100) and OOP (010) diffraction peaks. (d–f) Time evolution of UV–vis absorption contour in three active layers during spin coating. (g–i) Time evolution of the intensity of the main peak absorption and peak location of D‐polymer and NFA L8‐BO.

Despite this drawback in IP lamellar ordering, the strong intermolecular interactions introduced by the DTCPz‐SiO units helped maintain short π–π stacking distances in the OOP direction, partially mitigating the adverse effects of incomplete crystallization (Figure 3c). Interestingly, the overall OOP crystallinity was even enhanced with higher DTCPz‐SiO incorporation, with CCL values increasing from 20.0 Å for D18:L8‐BO to 22.8 Å for DN1:L8‐BO and 23.6 Å for DN2:L8‐BO. Notably, DN1 exhibited the most favorable molecular ordering for balanced charge transport, as reflected in its hole‐to‐electron mobility ratio (µ_h/µ_e) of 1.06, compared with 1.09 and 1.11 for D18:L8‐BO and DN2:L8‐BO, respectively. These charge transport characteristics, quantified via the space‐charge‐limited current (SCLC) method [50] (Figure S12 and Table S5), correlate well with the observed variations in FF across the corresponding OSC devices. Moreover, a noteworthy trend emerged when examining the temporal evolution of D and A phases during film formation. As the DTCPz‐SiO content increased from D18 to terpolymers, the onset of D‐polymer aggregation occurred increasingly earlier than that of the NFA (Figure 3g–i; Table S6), with the time gap widening from 35 ms for D18:L8‐BO to 67 ms for DN1:L8‐BO and 73 ms for DN2:L8‐BO. This clear increase in temporal offset is expected to promote more distinct phase separation and the formation of purer domains [47, 51, 52], likely driven by reduced D‐A miscibility. This interpretation is supported by calculations of the Flory‐Huggins interaction parameter (χ) [53] between each D‐polymer and L8‐BO (Figure S13 and Table S7), which showed progressively higher χ values with increasing DTCPz‐SiO content—0.22, 0.30, and 0.34 for D18:L8‐BO, DN1:L8‐BO, and DN2:L8‐BO, respectively, indicating weaker miscibility and stronger phase segregation.

Atomic force microscopy (AFM) was employed to characterize the surface morphology of the blend films. The results revealed a clear trend of increasing surface roughness with higher DTCPz‐SiO content, as indicated by larger root‐mean‐square (RMS) values of 1.57 nm, 1.60 nm, and 1.72 nm for D18:L8‐BO, DN1:L8‐BO, and DN2:L8‐BO, respectively (Figure 4a–c). Complementary AFM‐infrared (AFM‐IR) spectroscopy was conducted to quantify the nanoscale phase structure (Figure 4d–f). The D18:L8‐BO blend exhibited the narrowest and most uniform fibril width distribution, measuring 20.6 nm on average (Figures S14 and S15), attributable to its superior miscibility with the acceptor. However, such fine‐scale morphology may hinder efficient charge transport by limiting domain purity and percolation pathways. In contrast, the DN2:L8‐BO blend showed the widest and most heterogeneous fibril size distribution, at 31.4 nm, consistent with excessive acceptor aggregation and reduced D‐A interfacial contact. Notably, the DN1:L8‐BO blend presented an intermediate morphology, featuring moderately sized fibrils of 25.5 nm and a relatively narrow size distribution, suggesting an optimal structure for both exciton dissociation and charge transport. To further quantify phase separation and domain purity, grazing incidence small‐angle X‐ray scattering (GISAXS) measurements were performed (Figure 4g; Figure S16). The average pure domain size (2R_g ) and the intermixing domain size (X _DAB) were extracted using the Debye‐Anderson‐Brumberger (DAB) and fractal network models [54, 55, 56], respectively, and are summarized in Figure 4h and Table S8. Among the three blends, DN1 demonstrated the most favorable phase structure, with a significantly larger 2R_g (26.1 nm) and a smaller X _DAB (26.7 nm) compared to D18 (23.0 nm and 29.0 nm), indicating denser domain packing and enhanced phase separation. These results are consistent with the AFM and AFM‐IR analyses, reinforcing the conclusion that DN1 achieves a well‐balanced morphology conducive to efficient charge generation and transport.

FIGURE 4.

(a–c) AFM height and corresponding phase images, and (d–f) AFM‐IR images of D18:L8‐BO, DN1:L8‐BO, and DN2:L8‐BO blend films. (g) Intensity profiles along the q _r axis extracted from 2D GISAXS data for the respective blends. (h) Fitted average pure domain size 2R_g and intermixing domain size X _DAB.

The time‐dependent optimally tuned range‐separated hybrid DFT (TD‐OT‐DFT) analysis of D18:L8‐BO and DNX:L8‐BO interfaces reveals key differences in their excited‐state characteristics (Figure 5a; Table S9). In the DNX:L8‐BO complexes, the energy offset between the local excitation (LE) and singlet charge‐transfer (CT¹) states is reduced, ranging from 0.140 eV to 0.272 eV, compared to 0.257 eV to 0.299 eV in D18:L8‐BO. This smaller LE‐CT¹ gap facilitates more efficient population transfer from LE to CT¹ states. Furthermore, the energy separation between the singlet CT¹ and triplet CT³ states is larger in DNX:L8‐BO (0.244–0.343 eV) than in D18:L8‐BO (0.188–0.252 eV), suggesting a reduced likelihood of non‐radiative recombination via triplet formation. Collectively, these results indicate that incorporating the siloxane‐functionalized DTCPz‐SiO unit into D18 enhances charge‐transfer efficiency and may improve exciton dissociation dynamics at the D‐A interface.

FIGURE 5.

(a) LE and CT states at the L8‐BO/D18 interface (left) and L8‐BO/DN1 interface (right), as obtained from TD‐OT‐DFT (see Figure S17 for other D‐A interface conformations). (b,c) 2D TA maps of D18:L8‐BO and DN1:L8‐BO blend films, and (d,e) the corresponding TA spectra probed at different time delays. (f) EQE_EL spectra of D18:L8‐BO‐, DN1:L8‐BO‐ and DN2:L8‐BO‐based devices. Summary of (g) V _OC and (h) FF as a function of PCE for OSCs fabricated using non‐halogenated solvents (toluene, o‐xylene, and p‐xylene) with CS₂ (1:1, v/v). (i) J–V curves of DN1:L8‐BO:AITC‐based ternary OSCs fabricated from different solvents.

To investigate exciton dynamics and interfacial charge transfer processes, we employed transient absorption (TA) spectroscopy (Figure 5b–e; Figure S18). At the D‐A interface, a prompt decay of the L8‐BO ground‐state bleach (GSB) signal at 790 nm, accompanied by a concurrent rise in the D‐polymer GSB signal at 580 nm, clearly indicated efficient hole transfer from the A to the D. We extracted the temporal evolution of the donor GSB signal at 580 nm to monitor hole transfer kinetics, fitting it using a mono‐exponential function [57, 58, 59] (Figure S19). The resulting time constants (τ), corresponding to exciton dissociation dynamics at the interface or in mixed domains, were determined to be 0.959 ps for D18:L8‐BO, 0.610 ps for DN1:L8‐BO, and 0.732 ps for DN2:L8‐BO. These results confirm that DN1 exhibits the most efficient and rapid interfacial hole transfer process.

Next, EQE of electroluminescence (EQE_EL) was measured at a current density corresponding to the short‐circuit condition to reflect practical charge recombination behavior (Figure 5f) [60, 61]. It is directly correlated with nonradiative voltage losses, following the relation ΔV _OC ^nonrad = ‐(kT/q)ln(EQE_EL), where k is the Boltzmann constant, T is the temperature in Kelvin, and q is the elementary charge. For OSCs based on D18:L8‐BO, DN1:L8‐BO, and DN2:L8‐BO, the EQE_EL values were 1.86 × 10⁻⁴, 2.55 × 10⁻⁴, and 3.14 × 10⁻⁴, corresponding to ΔV _OC ^nonrad values of 0.222, 0.213, and 0.208 eV, respectively, indicating significantly suppressed nonradiative recombination in the terpolymer‐based systems. This suppression contributes to reduced ΔV _OC ^nonrad, and explains the observed variation in V _OC with composition.

The impact of the incorporation of the third component on charge generation and extraction was also analyzed via the relationship between photocurrent density (J _ph) and effective voltage (V _eff) [62]. As shown in Figure S20, the DN1:L8‐BO device exhibited superior exciton dissociation efficiency (P _diss) and charge collection probability (P _coll), with values increasing from 0.983/0.874 (for D18) and 0.974/0.858 (for DN2) to 0.988/0.887 (for DN1). This enhancement indicates that the optimized morphology of the DN1 blend promotes both exciton separation and carrier collection. Moreover, light‐intensity dependent measurements provided additional insights into recombination dynamics (Figure S21). For DN1:L8‐BO, the slope of V _OC versus ln(P _light) approaches the ideal value of kT/q, signifying minimized trap‐assisted recombination [63]. Simultaneously, the log(J _SC) versus log(P _light) plots yielded an α value close to 1 [64], reflecting suppressed bimolecular recombination. Transient photocurrent (TPC) and transient photovoltage (TPV) measurements (Figure S22) further confirmed improved charge dynamics in the DN1:L8‐BO blend. DN1 exhibited a markedly shortened charge extraction time of 0.453 µs and an extended carrier lifetime of 5.24 µs, indicating balanced, efficient charge transport and suppressed recombination losses [65]. Complementary capacitance‐frequency measurements and Gaussian fitting analysis (Figure S23) revealed a significant reduction in trap density (N _t) [66] to 3.27 × 10¹⁵ cm⁻³ for DN1:L8‐BO, supporting the reduction of trap‐assisted recombination pathways.

Therefore, the minor incorporation of the third component, DTCPz‐SiO in DN1 optimizes device performance by enhancing exciton dissociation, charge generation, and extraction processes, while simultaneously suppressing recombination and reducing nonradiative voltage losses. These synergistic effects collectively contribute to the superior FF, V _OC, and J _SC, enabling DN1‐based devices to outperform the parent D18 systems.

Beyond these performance enhancements, the terpolymer design also confers remarkable solvent versatility, a critical advantage for scalable and eco‐friendly fabrication. Specifically, DN1 demonstrates excellent solubility in both halogenated and non‐halogenated solvents, enabling high compatibility with diverse processing conditions (Figures S24 and S25). To further explore this versatility beyond the chloroform‐processed devices discussed earlier, we fabricated OSCs using non‐halogenated solvents within a conventional device architecture (ITO/PEDOT:PSS/D:L8‐BO/PNDIT‐F3N/Ag). The corresponding PV performance parameters are summarized in Figure S26 and Tables S10–S12. As shown in Figure 5g,h, the D18:L8‐BO device processed from a binary solvent system of o‐XY:CS₂ = 1:1 (v/v) achieves a maximum PCE of 18.1%, lower than its chloroform‐processed counterpart, primarily due to reductions in both V _OC and FF. In stark contrast, DN1:L8‐BO‐based devices consistently deliver average PCEs around 19.5% across all tested non‐halogenated solvents. In addition to the device‐level solvent tolerance discussed above, we further investigated the intrinsic film‐formation behaviors of D18 and DN1 in non‐halogenated solvents to elucidate the molecular origins of their disparate processing robustness. As shown in Figure S27, temperature‐dependent UV–vis absorption spectra collected in non‐halogenated solvents exhibit aggregation trends analogous to those observed in chlorobenzene, indicating that the fundamental aggregation mechanism is preserved across different solvent systems. Time‐resolved in‐situ absorption measurements further reveal that DN1 undergoes an earlier aggregation onset and a shorter aggregation duration than D18 during film formation in non‐halogenated solvents (Figures S28 and S29 and Table S13), in close agreement with the chloroform‐based kinetics discussed previously. Moreover, GIWAXS characterization of non‐halogenated‐processed neat and blend films (Figure S30 and Table S14) demonstrates closer π–π stacking distances, enlarged CCL, and a higher degree of face‐on orientation for DN1, particularly along the OOP (010) direction. Collectively, these results indicate that the backbone planarization and solubility enhancement induced by the DTCPz‐SiO units are intrinsic characteristics that persist in non‐halogenated solvent systems, thereby underpinning the superior morphology evolution and solvent tolerance of DN1‐based blends. Furthermore, the incorporation of a relatively wide‐bandgap acceptor, AITC [2, 67, 68], into the DN1:L8‐BO blend to form a ternary OSC further optimizes the device parameters, yielding PCEs of 20.9% and 20.0% under halogenated (chloroform) and non‐halogenated (o‐XY:CS₂ = 1:1 (v/v)) solvent processing, respectively (Figure 5i; Figure S31 and Table 1), stemming from AITC's cascaded energy level alignment with the host materials and robust morphology modulation capability for the corresponding blend film, as confirmed in our previous work [2, 67, 68]. Achieving efficiencies surpassing 20% positions these OSCs among the highest‐performing devices reported to date for non‐halogenated solvent processing. These findings clearly demonstrate that the DN1‐based blend exhibits excellent tolerance to the nature of processing solvents, making it a promising candidate for environmentally friendly fabrication of high‐performance organic photovoltaics.

3. Conclusion

We presented a rational terpolymer design strategy to enhance both the photovoltaic performance and processing versatility of conjugated D‐polymers for OSCs. By incorporating a siloxane‐functionalized, electron‐deficient, and conformationally rigid DTCPz‐SiO unit into the D18 backbone, the resulting terpolymer DN1 exhibited improved backbone planarity, deeper HOMO energy levels, and enhanced π–π stacking with preferential face‐on orientation. These molecular‐level improvements translated into favorable aggregation kinetics, increased crystallinity, and optimized phase separation in the active layer. As a result, DN1:L8‐BO‐based devices achieved a remarkable PCE of 20.1% under conventional processing conditions, while maintaining high efficiencies of ∼19.5% when processed from non‐halogenated solvents. The device performance is further improved under both halogenated and non‐halogenated conditions using DN1:L8‐BO:AITC ternary OSCs, which deliver PCEs of 20.9% and 20.0%, respectively, meeting industrial roll‐to‐roll manufacturing requirements. This work establishes siloxane‐tethered rigid building blocks as an effective molecular design motif for donor polymers, enabling the simultaneous optimization of efficiency, morphological control, and non‐halogenated solvent processability, and providing general design guidelines for next‐generation scalable and sustainable organic photovoltaic materials.

Conflicts of Interest

The authors declare no conflicts of interest.

Supporting information

Supporting File: adma73225‐sup‐0001‐SuppMat.docx.

Data Availability Statement

The data that support the findings of this study are available from the corresponding author upon reasonable request.