Photovoltage enhancement of M-series acceptor-based polymer solar cells and minimodules through the modulation of charge-transfer states

ABSTRACT

Side-chain fluorination can enhance the backbone organization and carrier mobility of non-fullerene acceptors (NFAs) but it often reduces their photovoltage due to the resulting deeper-lying lowest unoccupied molecular orbital (LUMO) levels. Herein, we present a strategy to regulate the LUMO levels of two NFAs, MC9F5 and MC7F5, by repositioning the highly electronegative –C₂F₅ moieties on the side chains. This approach mitigates the impact of fluorination on the energy levels, thereby improving the photovoltage and overall device performance. By incorporating 10,10,11,11,11-pentafluoro-2-(8,8,9,9,9-pentafluorononyl)undecyl side chains, the –C₂F₅ moieties are positioned away from the conjugated backbone of MC9F5, resulting in an elevated LUMO level compared with MC7F5, which features 8,8,9,9,9-pentafluoro-2-(6,6,7,7,7-pentafluoroheptyl)nonyl side chains. This modification reduces both the charge generation and the non-radiative energy losses in the MC9F5-based devices. The MC9F5-based small-area and minimodule devices achieve efficiencies of 18.02% and 15.66%, respectively, which are among the highest values reported for acceptor–donor–acceptor-type NFAs. This study highlights a valuable fluorination strategy for achieving high-performance NFAs.

INTRODUCTION

Organic solar cells (OSCs) represent a promising complementary photovoltaic technology to the currently dominant silicon-based solar cells, offering distinct advantages such as flexibility, light weight, solution processability and semitransparency [1–7]. These characteristics make OSCs particularly well suited for applications such as building-integrated photovoltaics, automotive components and various portable power sources for both indoor and outdoor use [8–12]. Recently, the power conversion efficiencies (PCEs) of OSCs have significantly improved due to the emergence of non-fullerene acceptors (NFAs), which feature long-wavelength absorption capabilities and minimized energy losses when blended with wide-bandgap copolymers [13–18]. The adoption of NFAs has markedly enhanced the photovoltaic performance of OSCs by increasing both photocurrent density and voltage output [13–17]. In 2019, a benchmark fused-ring NFA, Y6, with an acceptor–donor–acceptor–donor–acceptor (ADA'DA) configuration, was reported to have achieved an outstanding efficiency of 15.7% [18]. Since then, extensive efforts have been made by many research groups to chemically modify Y6, leading to the realization of several high-performance ADA'DA-type NFAs, such as L8-BO, N3 and BTP-eC9 [19–21]. To date, the majority of OSCs that demonstrate PCEs of >18% have predominantly employed Y6-series acceptors, whereas high-performance OSCs that utilize ADA-type NFAs remain comparatively scarce. Therefore, it is essential to develop new molecular design strategies to expand the range of high-performance NFA materials.

In 2020, our group reported a new class of acceptors known as M-series acceptors, which are based on an ADA-type architecture and synthesized by substituting all sp³-hybridized bridging carbon atoms in the ladder-type donor core with sp²-hybridized bridging nitrogen atoms [22–25]. The M-series acceptors demonstrate reduced synthetic complexity and exhibit a superior figure of merit compared with several representative NFAs, such as IT-4F and Y6 [26]. Furthermore, these linear-shaped ADA-type NFAs show extended light absorption and an ordered 3D network of molecular stacking, which is facilitated by enhanced intermolecular π–π interactions [27]. Initially, the M-series acceptors achieved a best PCE of 15.2%, which was subsequently improved to 17% through molecular structure optimization and morphology control [22,27]. More recently, we introduced partially fluorinated and branched side chains to enhance the backbone organization of the M-series acceptor (MC7F3) [28]. This modification facilitated the formation of a 3D network packing structure, thereby enhancing carrier transport and boosting the device efficiency to 17.61%. However, fluorinated side chains, while enhancing short-circuit current density (J_sc) and fill factor (FF), induce a downshift in the lowest unoccupied molecular orbital (LUMO) level, consequently lowering the charge-transfer (CT) state and open-circuit voltage (V_oc) of the devices. It remains challenging to leverage the positive effects induced by side-chain fluorination while mitigating their adverse effects.

In contrast to non-excitonic solar cells (e.g. inorganic semiconductor cells), in which excitons with low binding energies can be dissociated by thermal energy, excitonic OSCs involve the CT state due to the low dielectric constant of organic materials [29–32]. It has been disclosed that the CT state energy is approximately equal to the energy difference between the LUMO of the acceptor material and the highest occupied molecular orbital (HOMO) of the donor material [33]. The presence of a CT state results in an additional energy-loss channel, ΔE_CT, and a significant non-radiative energy loss, ΔE_non-rad. The ΔE_CT represents the energy loss associated with charge generation, linked to the essential driving force for exciton dissociation in OSCs. It can be calculated by using the equation of ΔE_CT = E_g − E_CT, where E_g is the optical bandgap of the donor/acceptor materials and E_CT is the energy of the CT state [33]. Thus, increasing E_CT by minimizing the energetic offset between the donor and acceptor materials can effectively reduce ΔE_CT [33,34]. Importantly, the reduction in ΔE_CT can enhance the hybridization between the highly luminescent localized excitation (LE) state and the CT state [35,36]. Through an intensity borrowing mechanism, this can increase the luminescence of the CT state, thus decreasing ΔE_non-rad [36]. However, the reduction of ΔE_CT would also increase the risk of insufficient exciton dissociation and charge generation, leading to a decrease in external quantum efficiency (EQE) [37–39]. Thus, the precise modulation of E_CT in OSCs is fundamentally important for energy-loss reduction and device performance optimization.

Herein, we strategically designed and synthesized two partially fluorinated and branched side chains, namely 8,8,9,9,9-pentafluoro-2-(6,6,7,7,7-pentafluoroheptyl)nonyl and 10,10,11,11,11-pentafluoro-2-(8,8,9,9,9-pentafluorononyl)undecyl groups, which were incorporated into the M-series acceptor framework, resulting in two new NFAs (MC7F5 and MC9F5 in Fig. 1). Moving the strongly electronegative –C₂F₅ terminal groups away from the conjugated skeleton leads to a lower surface electrostatic potential and an elevated LUMO energy level for MC9F5 in comparison with those for MC7F5. This chemical modification elevates the CT state and reduces the gap between E_g and E_CT, effectively lowering both ΔE_CT and ΔE_non-rad, which contributes to an increased V_oc in the MC9F5-based devices. Moreover, the longer side chains in MC9F5 suppressed molecular aggregation, promoting a more favorable blend film morphology and an improved FF. As a result, the optimized MC9F5-based OSC achieved an excellent PCE of 18.02%, along with a high FF of 80.14% and a V_oc of 0.878 V. In comparison, the MC7F5-based OSC showed a lower PCE of 17.20%, with an FF of 78.85% and a V_oc of 0.832 V. Notably, both the 18.02% PCE and the 80.14% FF are among the highest values reported for OSCs that utilize ADA-type NFAs. Furthermore, MC9F5 exhibited excellent film-forming properties, making it highly suitable for large-area OSC fabrication. Encouragingly, the MC9F5-based minimodule device with an effective area of 11.09 cm² showed an outstanding PCE of 15.66%, which is comparable to those of the modules that utilize Y-series acceptors (Table S1). To the best of our knowledge, MC9F5 is the first non-Y6 acceptor to achieve a PCE of >15% in a module device, demonstrating its significant potential for practical applications.

RESULTS AND DISCUSSION

The two acceptors, MC7F5 and MC9F5, designed in this study (Fig. 1a), were synthesized by following our previously reported method [40,41], as illustrated in Schemes S1 and S2. Their molecular structures were characterized via ¹H nuclear magnetic resonance (NMR), ¹⁹F NMR and high-resolution mass spectroscopy, with further details provided in the Supplementary Data. Notably, the presence of neighboring side chains in both acceptors imparts exceptional solubility in halogenated and aromatic solvents, such as dichloromethane, chloroform, toluene and chlorobenzene. The result demonstrates that a low F/H atomic ratio on the alkyl chain does not compromise the solubility of the acceptors in regular solvents, ensuring the successful material purification and device fabrication. To elucidate how the position of the –C₂F₅ groups on the side chain affects the molecular geometry and electronic properties, density functional theory (DFT) calculations were performed on the two NFAs at the B3LYP/6–31G** level. As illustrated in Fig. S1, both NFAs exhibit a high degree of planarity, which is conducive to achieving well-ordered intermolecular assembly and hence the intermolecular electron transport. Except for the carbonyl, cyano and halogen groups, MC7F5 and MC9F5 exhibit similar positive electrostatic potential (ESP) distributions across their π-conjugated backbones, suggesting their typical electron-accepting characteristics (Fig. 1b). Figure 1c presents the average ESP values for each atom within the fused-ring core of the two NFAs. It can be observed that the ESP values of all atoms in the fused-ring core decrease as the –C₂F₅ motifs move away from the conjugated backbone. This trend reveals an inverse relationship between the distance and the impact of the highly electronegative –C₂F₅ group on the electron density distribution across the conjugated backbone. The change in electron density distribution would further affect the electronic energy levels of the NFAs. As shown in Fig. S2, although the spatial distributions of the frontier molecular orbitals are quite similar for the two NFAs, their HOMO and LUMO energy levels are different. The calculated LUMO and HOMO energy levels for MC7F5 are –3.57 and –5.66 eV, respectively, whereas those for MC9F5 were shifted upward to –3.54 and –5.62 eV, respectively. This shift is attributed to the highly electronegative fluorine atoms that lower the charge density on the conjugated backbone through electrostatic interactions. These findings indicate that the ESP distribution and frontier orbital energy levels of NFAs can be finely tuned by varying the distance between the electronegative –C₂F₅ group and the π-conjugated backbone.

The absorption spectra of the two acceptors were measured and compared in both diluted solutions and thin films, and the detailed photophysical data are summarized in Table S2. Due to the same molecular backbone, MC7F5 and MC9F5 exhibit similar absorption profiles in dilute chloroform solution (Fig. 2a). The absorption peak of MC7F5 is observed at 731 nm, demonstrating a blue shift of ∼4 nm in comparison with MC9F5. Such a shift is attributed to the shorter distance between the –C₂F₅ moieties and the molecular backbone in MC7F5 compared with that in MC9F5, which reduces the electron-donating character of the fused-ring unit in MC7F5, thereby weakening the intramolecular CT effect, as evidenced by the theoretical calculation results. The maximum extinction coefficients (ε_max) for MC7F5 and MC9F5 in solution were determined to be 1.98 × 10⁵ and 2.13 × 10⁵ M^–1 cm^–1, respectively. The transition from solution to solid state results in bathochromic shifts in the absorption spectra of both NFAs—a phenomenon that suggests the formation of stronger intermolecular π–π interactions in the solid films. Interestingly, MC7F5 is blue-shifted in solution but red-shifted by 2 nm in solid film compared with MC9F5, which indicates stronger molecular stacking of MC7F5 than MC9F5 in the solid state. The MC9F5 film exhibits a stronger absorption band within the range of 700–900 nm, with an absorption coefficient of 2.82 × 10⁵ cm^–1, surpassing that of the MC7F5 film, which has an absorption coefficient of 2.42 × 10⁵ cm^–1. The optical band gaps (E_g) of MC7F5 and MC9F5, determined from thin-film absorption edges, were measured at 1.36 and 1.38 eV, respectively, with both values closely approaching the ideal bandgap (1.34 eV) as defined by the Shockley–Queisser limit [42].

Cyclic voltammetry (CV) measurements were performed to investigate the electrochemical properties of MC7F5 and MC9F5, and the relevant data are summarized in Table S2. Based on the CV curves that are presented in Fig. 2c and Fig. S3, the HOMO and LUMO energy levels of MC7F5 were determined to be –5.58 and –4.07 eV, respectively. In comparison, MC9F5 exhibited slightly higher energy levels at –5.52 eV (HOMO) and –4.02 eV (LUMO), as illustrated in Fig. 2d. The variation in energy levels agrees well with the DFT calculations. Both the HOMO and LUMO energy levels of MC7F5 are downshifted compared with those of MC9F5. This phenomenon can be primarily attributed to the shorter regular alkyl chain distance between the –C₂F₅ moieties and the molecular backbone, which enhances electrostatic interactions and consequently reduces the electron-richness of the fused-ring core in MC7F5. Notably, reduced energy levels in the acceptor molecule would lead to increased energy offsets relative to the donor material, thereby promoting exciton dissociation and potentially enhancing J_sc. However, this also leads to in an increased E_loss in the resulting OSCs.

The crystalline properties of the neat NFA films were then investigated by using grazing-incidence wide-angle X-ray scattering (GIWAXS). The 2D GIWAXS patterns, along with the corresponding line cuts, are presented in Fig. S4 and the results of the quantitative analysis are summarized in Table S3. As illustrated in the figure, despite the minimal difference of only two carbon atoms in their alkyl chain lengths, the two NFAs exhibit remarkably distinct molecular packing arrangements in the solid state. The 2D pattern of MC7F5 exhibits well-defined diffraction spots with high intensity, suggesting pronounced molecular aggregation characteristics. These structural features align well with the optical properties observed in the thin-film absorption measurements. Notably, we observed that the diffraction peaks located in the out-of-plane direction are split, which might be due to the polycrystalline texture of MC7F5. For the neat MC7F5 film, the (010) π–π diffraction peak is primarily located along the q_z-axis, whereas the (100) lamellar peaks appear along the q_xy-axis, indicating a preferred face-on orientation relative to the substrate. In contrast, the MC9F5 crystallites exhibit edge-on-dominated orientation, as indicated by the π–π diffraction peak along the q_xy-axis and the strong (100) lamellar diffraction peak present along both the q_xy- and q_z-axes. Furthermore, unlike the MC7F5 film, which shows multiple narrow diffraction arcs in the out-of-plane direction beyond q_z = 0.7 Å^–1, the MC9F5 film reveals some diffraction points at azimuthal angles between 30^o and 70^o. These diffraction points are distributed similarly to the ‘satellite points’ that surround the main (100) peak, suggesting the formation of a distinct mixed polycrystalline texture with small crystalline domains [43]. The results show that, in addition to tuning the energy levels, extension of the partially fluorinated side chains can effectively restrain the excessive molecular aggregation while ensuring ordered molecular stacking, promoting the formation of a favorable nanoscale blend film morphology.

To assess the photovoltaic performance of MC7F5 and MC9F5, OSCs were fabricated by using a conventional device architecture of indium tin oxide (ITO)/[2-(9H-carbazol-9-yl)ethyl] phosphonic acid (2PACz)/active layer/2,9-bis(3-(dimethylamino)propyl)anthra[2,1,9-def:6,5,10-d′e′f′]diisoquinoline-1,3,8,10(2H,9H)-tetraone (PDIN)/Ag. In this work, 2PACz was selected as the hole transport layer to minimize absorption losses and reduce contact resistance [44]. The wide-bandgap polymer PM6 (Fig. S5), which has a low-lying HOMO energy level of –5.48 eV and an absorption range from 300 to 680 nm, was chosen as the electron donor material due to its complementary absorption profile and well-matched energy levels with the two NFAs (Fig. 2d) [45]. The active layer was spin-coated from a chlorobenzene solution with chloronaphthalene (CN) as an additive. The device parameters under varying fabrication conditions are summarized in Supplementary Tables S4 and S5. The J–V curves and the relevant photovoltaic parameters for the optimized OSCs are provided in Fig. 3a and Table 1, respectively. The optimized device based on PM6:MC7F5 delivered a PCE of 17.20% with a V_oc of 0.832 V, a high J_sc of 26.22 mA cm^–2 and an FF of 78.85%. To our delight, a higher PCE of 18.02% was obtained from the optimized PM6:MC9F5 binary device with a higher V_oc of 0.878 V, a J_sc of 25.62 mA cm^–2 and an FF of 80.14%. Notably, the PCE of 18.02% for the MC9F5-based device is one of the highest values for ADA-type NFAs reported so far (Fig. S6 and Table S6). To further assess the compatibility of MC9F5 with other donor materials, we selected two commonly used polymer donors, PM1 and D18 [28], for a comprehensive investigation (Table S7). The PCEs of the resulting OSCs based on PM1:MC9F5 and D18: MC9F5 blends were 15.10% and 16.01%, respectively, both of which are lower than that of the PM6:MC9F5-based OSC. The improved photovoltaic performance of the PM6:MC9F5-based OSC in comparison with that based on PM6:MC7F5 is primarily due to its higher V_oc and FF values. This enhancement stems from the elevated CT state that reduces the E_loss within the PM6:MC9F5 device and a more balanced charge transport, as detailed in the following discussion. For comparison, we also used MC7F0 (Fig. S7), which has the regular non-fluorinated side chains, as the acceptor material to fabricate the OSCs [28]. The MC7F0-based device exhibited a V_oc of 0.88 V, which is higher than that of the MC7F5-based device but very close to that of the MC9F5-based device. This suggests that the higher E_loss caused by incorporating fluorinated side chains can be effectively mitigated by the strategy of moving the fluorinated groups on the side chains away from the conjugated backbone.

The MC9F5-based devices showed remarkable FF values that are on a par with those of state-of-the-art OSCs, reflecting optimized charge transport and suppressed recombination. The maximum FF (FF_max) without charge-transport loss can be calculated empirically by using the following equation: |$F{F_{max}} = \frac{{{\gamma _{oc}} - {\rm{l}}n( {{\gamma _{oc}} + 0.72} )}}{{{\gamma _{oc}} + 1}}$|⁠, where γ_oc = qV_oc/nkT, q is the elementary charge, n is the diode ideality factor, k is the Boltzmann constant and T is the temperature [46]. As shown in Fig. 3b, the champion FF of the optimized device reaches 80.14%, which corresponds to 92% of the FF_max of OSCs with a V_oc of 0.878 V. Figure 3b and Table S6 display the dependence of the FF_max on V_oc, along with the FF values of the optimal devices from this study and previously reported OSCs based on ADA-type NFAs. Notably, the PM6:MC9F5 device achieved a higher FF value compared with other reported OSCs that utilize ADA-type NFAs. The long-term storage stability of the unencapsulated PM6:MC7F5- and PM6:MC9F5-based devices was assessed in a glove box at room temperature. As shown in Fig. S8, after ∼600 h of storage, both devices could maintain >98% of their initial PCEs, indicating a good storage stability.

Figure 3c illustrates the EQE response spectra for the two best-performing OSCs. Compared with the PM6:MC9F5-based device, the PM6:MC7F5-based device exhibited a broader photoresponse range and slightly higher EQE values, which contribute to its increased J_sc value. The integrated current densities that were derived from the EQE curves for the PM6:MC7F5 and PM6:MC9F5 devices are 24.96 and 24.38 mA cm^–2, respectively, aligning closely with the J_sc values that were obtained from the J–V measurements with deviations within 5.0%. In bulk-heterojunction (BHJ) OSCs, the photon-to-electron conversion efficiency relies heavily on exciton dissociation. Accordingly, we analysed the charge dissociation probability (P_diss) for the two best-performing OSCs by examining the relationship between the photocurrent density (J_ph) and the effective voltage (V_eff) (Fig. S9). The PM6:MC7F5-based device exhibits a higher P_diss value of 98.90% compared with 97.17% for the PM6:MC9F5-based counterpart, demonstrating enhanced exciton dissociation efficiency. This increased charge dissociation probability directly correlates with enhanced current generation in the PM6:MC7F5-based device.

The recombination kinetics in OSCs were quantitatively analysed by evaluating the dependence of J_sc on the illumination intensity (P_light). The relationship between J_sc and P_light can be given by using the following equation: J_sc ∝ P_light^α, where α is the exponential factor. The α values for the PM6:MC7F5- and PM6:MC9F5-based devices, determined through linear fitting, were found to be 0.96 and 0.98, respectively (Fig. 3d). The α value closer to 1 for the PM6:MC9F5-device suggests a lower degree of charge recombination within this device. Space-charge-limited current measurements were then conducted to evaluate the charge-transport properties of the PM6:MC7F5 and PM6:MC9F5 blends. The J–V curves of the hole-only and electron-only devices under dark conditions are depicted in Fig. S10 and the calculated hole mobility (μ_h) and electron mobility (μ_e) are listed in Table S8. The μ_h and μ_e of the PM6:MC7F5 blend film were determined to be 1.09 × 10^–3 and 5.11 × 10^–4 cm² V^–1 s^–1, respectively, whereas those of the PM6:MC9F5 blend film were 7.98 × 10^–4 and 4.33 × 10^–4 cm² V^–1 s^–1, respectively. Consequently, the PM6:MC9F5 blend film achieved a more balanced μ_h/μ_e ratio of 1.84, compared with the ratio of 2.13 for the PM6:MC7F5 blend film. More balanced charge transport and suppressed charge recombination in the PM6:MC9F5-based devices contribute to their higher FFs.

Given that the bulky partially fluorinated side chains can effectively inhibit severe molecular aggregation, we chose to use MC9F5 as the acceptor material for the fabrication of large-area devices. A photograph of the minimodule device, which consists of four subcells that are connected in series, is shown in Fig. 3e. The corresponding schematic diagram of the device structure is illustrated in Fig. S11. The measured I–V curve of the minimodule and the corresponding photovoltaic parameters are provided in Fig. 3f. This 11.09-cm² module yielded a V_oc of 3.56 V, an I_sc of 67.46 mA, an FF of 72.37% and a PCE of 15.66% with respect to the active area. To the best of our knowledge, this is the first report of a module device based on a non-Y-series acceptor with an efficiency of >15% (Table S1), demonstrating the great potential of MC9F5 for fabricating large-area OSCs. Furthermore, the MC9F5-based minimodule device also exhibits good shelf stability with a low PCE loss of ∼6% after >1100 h of storage in an N₂-filled glovebox (Fig. S12).

To investigate the difference in V_oc values between the two OSCs, we conducted a detailed analysis of the E_loss of the devices by measuring their Fourier-transform photocurrent spectroscopy external quantum efficiency (FTPS-EQE) and electroluminescence (EL) spectra. Generally, the E_loss in OSCs can be divided into three parts, as represented by the equation: E_loss = ΔE₁ + ΔE₂ + ΔE₃ = (E_g – |$qV_{{\rm{oc}}}^{{\rm{SQ}}}$|⁠) + (⁠|$qV_{{\rm{oc}}}^{{\rm{SQ}}} - qV_{{\rm{oc}}}^{{\rm{rad}}}$|⁠) + (⁠|$qV_{{\rm{oc}}}^{{\rm{rad}}} - q{V_{{\rm{oc}}}}$|⁠), where |$qV_{{\rm{oc}}}^{{\rm{SQ}}}$| is the possible V_oc under the Shockley–Queisser limit and |$qV_{{\rm{oc}}}^{{\rm{rad}}}$| represents the V_oc when considering only the radiative recombination [47]. ΔE₁ is attributed to the radiative energy loss above the band gap, which is unavoidable for any solar cells. Here, the ΔE₁ values for both the PM6:MC7F5- and the PM6:MC9F5-based devices are ∼0.263 eV (Table S9). The second component ΔE₂ is the radiative energy loss below the band gap, which is correlated with the energy difference (ΔE_CT) between the bandgap (E_g^PV) and the charge-transfer state (E_CT) of the OSC devices. The E_g^PV values were 1.45 and 1.47 eV for the MC7F5- and MC9F5-based devices, respectively, as determined by using the cross point of the normalized emission and the absorption of the neat acceptor films (Fig. S13). The E_CT values of the MC7F5- and MC9F5-based devices can be obtained by fitting FTPS-EQE and EL spectra. As shown in Fig. 4a and b, the E_CT of the MC7F5-based OSC is determined to be 1.38 eV, whereas that of the MC9F5-based OSC is 1.44 eV. Correspondingly, the ΔE_CT value decreases from 0.07 to 0.03 eV, which is likely to be the primary factor that leads to the relatively lower P_diss and reduced J_sc observed in the MC9F5-based device. Moreover, changes in the ΔE_CT values also lead to variations in ΔE₂, as presented in Fig. 4d, in which the ΔE₂ values for the MC7F5- and MC9F5-based devices are 0.086 and 0.052 eV, respectively. ΔE₃ is the non-radiative recombination loss and can be quantified as (–kT/q)ln(EQE_EL), where k is the Boltzmann constant, T is the Kelvin temperature and EQE_EL is the electroluminescence quantum efficiency of the device. It is generally assumed that lowering the ΔE_CT of the OSC facilitates hybridization between the LE and CT states, which can enhance the luminescence of the CT state by borrowing the highly luminescent LE state [36]. Also, a smaller ΔE_CT facilitates the transition from the CT state back to the LE state, thus allowing the CT state to undergo radiative recombination through the emissive LE state. In any case, enhancing the luminescence of the CT state can effectively reduce the non-radiative recombination loss (i.e. ΔE₃) of the OSC. To confirm this mechanism and quantitatively calculate the ΔE₃ value, the EQE_EL spectra of the two OSCs were measured. As shown in Fig. 4c, the MC7F5-based device exhibits an EQE_EL of 9.19 × 10^–6, corresponding to a ΔE₃ of 0.290 eV whereas the MC9F5-based device shows a higher EQE_EL of 3.17 × 10^–5, leading to a lower ΔE₃ of 0.259 eV. Based on the above data, the E_loss of the MC9F5-based device is calculated to be 0.576 eV, which is considerably lower than the E_loss of 0.639 eV observed for the MC7F5-based device. The results indicate that the significant improvement in V_oc of the MC9F5 device can be attributed to the elevated E_CT induced by the increased LUMO energy level, which leads to the simultaneous reduction of radiative and non-radiative recombination losses.

The top-surface morphology of the MC7F5- and MC9F5-based BHJ blend films was examined by using atomic force microscopy (AFM) in a peak force quantitative nanomechanical mapping mode. As shown in Fig. 5a and b, both blend films display a fibrillar network morphology, which is beneficial for efficient charge transport. The root-mean-square roughness values under optimal conditions are 2.10 nm for the MC7F5-based film and 1.75 nm for the MC9F5-based film. The higher roughness value in the MC7F5-based film is attributed to its stronger molecular aggregation. AFM phase images (Fig. 5c and d) reveal phase separation in both blends; however, the PM6:MC7F5 blend film shows more pronounced phase separation and relatively larger phase domains compared with the PM6:MC9F5 blend film. The larger phase domains in the PM6:MC7F5 blend film may promote charge recombination, thereby adversely affecting the FF of the OSC devices. GIWAXS was subsequently used to analyse the microstructures and molecular packing behaviors of the blend films. As shown in Fig. 5e and f, both blend films exhibit distinct π–π stacking diffraction peaks along the q_z-axis and lamellar stacking diffraction peaks along the q_xy-axis, indicating the formation of a preferential face-on orientation. For the π–π stacking diffraction in the q_z-axis, the PM6:MC9F5 blend film exhibits a strong diffraction peak at 1.73 Å^–1, corresponding to a π–π stacking distance (d_π–π) of 3.63 Å, with a coherent length (CL) of 16.44 Å (Table S3). In comparison, the (010) diffraction peak of the PM6:MC7F5 blend film is slightly shifted to 1.75 Å^–1, giving a slightly shorter d_π–π of 3.59 Å with an enhanced CL of 18.85 Å. The enhanced crystallinity of the PM6:MC7F5 blend film is favorable for achieving a higher J_sc value in the resulting devices.

CONCLUSION

In conclusion, we have successfully designed, synthesized and characterized two M-series NFAs, namely MC7F5 with 8,8,9,9,9-pentafluoro-2-(6,6,7,7,7-pentafluoroheptyl)nonyl side chains and MC9F5 with 10,10,11,11,11-pentafluoro-2-(8,8,9,9,9-pentafluorononyl)undecyl side chains, for applications in OSCs. Extending the length of the partially fluorinated and branched side chains in MC9F5 effectively suppresses molecular aggregation, resulting in blend films with an optimal nanoscale phase separation morphology that is characterized by small domain sizes and good crystallinity. This favorable morphology facilitates balanced charge transport and reduces bimolecular recombination, yielding an impressive FF of 80.14% in the MC9F5-based devices, surpassing that of the MC7F5-based devices. Furthermore, positioning the fluorinated –C₂F₅ groups farther from the conjugated backbone diminishes the impact of the strong electronegativity of the fluorine atoms on the electronic properties of the resulting MC9F5, raising its LUMO energy level as well as the E_CT of the resulting blend film. The elevated E_CT reduces both radiative and non-radiative recombination losses, thereby improving the V_oc of MC9F5-based devices. Consequently, the optimized OSC based on PM6:MC9F5 achieved a remarkable PCE of 18.02%, which is one of the highest reported values for ADA-type fused-ring NFAs. In contrast, the PM6:MC7F5-based device delivered a moderate PCE of 17.20%, limited by the lower V_oc and FF values. Notably, MC9F5, with its favorable aggregation property, is also suitable for large-area device fabrication, achieving an excellent PCE of 15.66% in a minimodule device with an active area of 11.09 cm². Our findings demonstrate that fine-tuning the spatial separation between the fluorinated moieties of the side chain and the conjugated backbone allows precise modulation of both the LUMO energy levels and the E_CT characteristics, while maintaining optimal blend film morphology. This molecular engineering approach offers a promising design strategy for developing high-performance ADA-type acceptor materials.

FUNDING

This work was supported by the National Natural Science Foundation of China (52130306 and 22075287), the Self-deployment Project Research Program of Haixi Institutes, the Chinese Academy of Sciences (CXZX-2022-GH09 and CXZX-2022-JQ10) and the Program of Youth Innovation Promotion Association, Chinese Academy of Sciences (2021299).

AUTHOR CONTRIBUTIONS

Y.M. and Q.Z. proposed and supervised the project, as well as conceived and designed the experiments. K.X. conducted the synthesis and performed the majority of the structural characterizations and electrochemical tests. D.C. fabricated the OSC devices, and characterized and analysed the film morphology. J.-Y.W. performed theoretical calculations. D.W. and D.C. tested and analysed the energy loss. C.T. performed the GIWAXS characterizations. K.X., Y.M. and Q.Z. co-wrote the manuscript. All authors discussed the results and participated in analysing the experimental results.

Conflict of interest statement. None declared.

REFERENCES