Revealing the Critical Role of the HOMO Alignment on Maximizing Current Extraction and Suppressing Energy Loss in Organic Solar Cells.

For state-of-the-art organic solar cells (OSCs) consisting of a large-bandgap polymer donor and a near-infrared (NIR) molecular acceptor, the control of the HOMO offset is the key to simultaneously achieve small energy loss (Eloss) and high photocurrent. However, the relationship between HOMO offsets and the efficiency for hole separation is quite elusive so far, which requires a comprehensive understanding on how small the driving force can effectively perform the charge separation while obtaining a high photovoltage to ensure high OSC performance. By designing a new family of ZITI-X NIR acceptors (X = S, C, N) with a high structural similarity and matching them with polymer donor J71 forming reduced HOMO offsets, we systematically investigated and established the relationship among the photovoltaic performance, energy loss, and hole-transfer kinetics. We achieved the highest PCEavgs of 14.05 ± 0.21% in a ternary system (J71:ZITI-C:ZITI-N) that best optimize the balance between driving force and energy loss.

Introduction

Organic solar cells (OSCs) with a bulk heterojunction (BHJ) structure that respond to a wide solar spectrum and collect carriers with low energy loss have attracted significant attention owing to the quick improvement in device efficiency (Yu et al., 1995, Lu et al., 2015, Kang et al., 2016). The core issue is how to simultaneously achieve small energy loss (Vandewal et al., 2009, Yao et al., 2015) [Eloss = Egap-qVoc, where Egap is the optical bandgap, Voc is the open-circuit voltage and q is elementary charge.] and keep a high photocurrent. Recent progress shows success of using non-fullerene acceptors to extend thin-film absorption, while keeping a low energy loss, achieving a balanced trade-off between Voc and Jsc (short-circuit current) to enhance power conversion efficiency (Qian et al., 2018, Menke et al., 2018, Sun et al., 2018, Zhang et al., 2018a, Zhang et al., 2018b, Zhang et al., 2018c, Xie et al., 2018, Nian et al., 2018, Fei et al., 2018, Yuan et al., 2019, Yu et al., 2019, Liu et al., 2019, Cui et al., 2019). D-A-type non-fullerene acceptors (NFAs) have shown distinct advantages of highly tunable absorption and molecular energy level (Zhang et al., 2018a, Zhang et al., 2018b, Zhang et al., 2018c, Yan et al., 2018, Wadsworth et al., 2019, Cheng et al., 2018), which also exhibit efficient charge separation at a low driving force (Liu et al., 2016), a unique opportunity to minimize the Voc-Jsc trade-off. Within such scenario, maximum power conversion efficiency (PCE) for NFA-based single-junction OSCs is predicted to be 19% with an absorption onset at ∼860 nm and a low non-radiative recombination loss of ∼0.21 V (Hou et al., 2018).

State-of-the-art NFA-based OSCs contain wide/medium-bandgap n class="Species">donors and near-infrared (NIR) NFAs forming a complementary absorption. NIR NFAs are normally designed by introducing strong electron-accepting terminals or electron-donating bulky core to enhance intramolecular charge transfer property (Li et al., 2018a, Li et al., 2018b, Kan et al., 2017, Kan et al., 2018, Liu et al., 2017). A successful example can be seen from IT-4F (Zhao et al., 2017), which shows downshifted lowest unoccupied molecular orbital (LUMO) energy level and redshifted absorption compared with ITIC (Lin et al., 2015) owing to the addition of INCN-2F onto IDTT core. By matching with polymer donor PBDB-TF, IT-4F-based devices (Li et al., 2018b) show a higher average power conversion efficiency (PCEavg) of 13.30 ± 0.20%, however, with an undesirably higher Eloss, 0.66 eV, compared with ITIC-based devices (Yu et al., 2018) with a PCEavg of 10.0 ± 0.1% and an Eloss of 0.58 eV. Such dilemma is common in NIR NFA materials design. The bandgap that is associated with Jsc is on the contrary to energy loss reduction. The calculation of energy loss is based on NFA components featuring narrower optical bandgaps, and the Eloss can be reduced via the proper highest occupied molecular orbital (HOMO) regulation (Yao et al., 2016) at the stake of reducing the driving force for hole transfer defined by the energy difference between the optical bandgap of NIR NFAs and the energy of the charge transfer (CT) state in the blends (EgapNFA-ECT). A good demonstration can be seen from the comparison between BT-IC and BDT-IC (Kan et al., 2017, Li et al., 2017a, Li et al., 2017b). BT-IC shows a higher HOMO energy level than BDT-IC, −5.32 eV versus −5.51 eV. By matching a wide-bandgap polymer J71, BT-IC-based OSCs delivered a much lower Eloss of 0.53 eV than BDT-IC (Eloss: 0.61 eV). Currently, Yan et al. reported very low energy offsets between PffBT2T-TT donor and O-IDTBR acceptor delivering a maximum PCE of 10.4% (Chen et al., 2018) but only with a moderate maximum external quantum efficiency (EQEmax) of 67%. Thus, sufficient driving force should be guaranteed to ensure efficient charge separation to achieve the best OSC performance. However, as indicated in Figure S10 and Table S8 that summarized the EQEmaxs in NIR region against HOMO offsets for the reported OSCs with efficiencies over 12%, the relationship between HOMO offsets and the efficiency for hole separation is quite elusive so far, which can be attributed to the difficulty in decoupling the impacts originating from the blend morphology. To achieve the most optimized materials combination and the best device performance, we need a comprehensive understanding on how small a driving force can effectively perform the charge separation while obtaining a high photovoltage to ensure high OSC performance (Cha et al., 2018), a picture that is of particular importance to guide materials design to ensure low Eloss and high EQE, which is highly challenging without an ideal research model system and is the target of this work.

When designing A-D-A-type NFAs, the bulky electron-donating core based on fused heteroarenes determines the HOMO energy level. The bridging heteroatoms are effective to modulate the electron-donating ability of fused heteroarenes in terms of their various electronegativity, aromaticity, and σ-π conjugation effect (Yamaguchi and Tamao, 2015). Such protocol enables us to fine-tune the energy level of NIR NFAs. We recently developed indenoindene-based NIR acceptors NITI (Xu et al., 2017) and ZITI (Liu et al., 2018a) that exhibit high PCEs over 13% (Zhou et al., 2018). In this work, we manipulated NIR NFAs synthesized by introducing heteroatom bridge, yielding ZITI-S, ZITI-C, and ZITI-N, with precise control over HOMO energy levels and NIR absorptions. Theoretical calculation is useful to predict the variation trend of photoelectric properties in conjugated materials. Density functional theory (DFT) calculations (Figure 1) show that the backbone introduction of heteroatoms is an effective method in modulating HOMO energy level, whereas the LUMO energy level depends more on acceptor moieties. Despite the reduced optical bandgap from ZITI-S (1.61 eV), ZITI-C (1.47 eV) to ZITI-N (1.41 eV), J71:ZITI-S-, J71:ZITI-C-, and J71:ZITI-N-based solar cells show gradually increased Vocs of 0.812 ± 0.004, 0.851 ± 0.006, and 0.873 ± 0.005 V with Elosss of 0.80, 0.62, and 0.53 eV. Such materials provide a good platform to investigate the detailed correlations among Eloss, EQE, and driving force. We show that a moderate driving force down to ∼100 meV can ensure efficient charge transfer to achieve high short-circuit current. Such value in couple with energy loss pictures the fundamental physics of charge separation and energy level alignment at donor-acceptor interfaces. Detailed manipulation of hole transfer energy levels of combining double acceptor BHJ blends yielded a high PCE of 14.05 ± 0.21%. Thus, minimizing trade-off between charge transfer driving force and short-circuit current is an effective methodology that needs to be highlighted in both materials design and device optimization to ensure low Eloss and high PCE.

Figure 1

Chemical Structures and Energy-Level Alignment

(A) The molecular structures of IIDT-X and ZITI-X.

(B) Energy-level evolution of heteroarenes IIDT-S, IIDT-C, and IIDT-N by DFT calculations at B3LYP/6-31G** level.

(C) Optimized geometries and contour plots of frontier molecular orbitals with HOMO and LUMO energy levels of ZITI-X NFAs.

Chemical Structures and Energy-Levn class="Chemical">el Alignment

(A) The molecular structures of n class="Chemical">IIDT-X and ZITI-X.

(B) Energy-level evolution of heteroarenes IIDT-S, IIDT-C, and IIDT-N by DFT calculations at B3LYP/6-31G** levn class="Chemical">el.

(C) Optimized geometries and contour plots of frontier molecular orbitals with HOMO and LUMO energy levels of n class="Chemical">ZITI-X NFAs.

Results and Discussion

Materials Design and Photoelectric Property

Figure 1 shows the structures and DFT calculations of heteroatom bridged IIDTs. It is seen that the frontier energy levels of IIDT-X can be well controlled by selecting S, C, N bridge atoms. The HOMO of the resulting NFAs is located at IIDT-X center, and the LUMO is located at INCN-2F moieties. Thus, gradually changed HOMO energy levels and similar LUMO energy levels are expected for ZITI-X series. Detailed synthesis for ZITI-S, ZITI-C, and ZITI-N is shown in Supplemental Information. These acceptors show good thermal stability with high thermal decomposition temperatures of 318°C, 321°C, and 323°C for ZITI-S, ZITI-C, and ZITI-N, respectively (see Figure S1). The photophysical and electrochemical properties of ZITI-X acceptors are examined. The absorption spectra of J71 donor and ZITI-X acceptors in thin films are shown in Figure 2A. ZITI-S, ZITI-C, and ZITI-N exhibit maximum absorptions at 677, 717, and 747 nm with high molar extinction coefficients (ε) of 2.11 × 105, 2.46 × 105, and 2.90 × 105 M−1 cm−1 in solution, respectively (see Figure S2), and broad NIR absorptions between 500 and 900 nm with absorption peaks at 709, 756, and 798 nm in thin films that are redshifted by 32, 39, and 51 nm. The optical bandgaps of ZITI-S, ZITI-C, and ZITI-N are estimated to be 1.61, 1.47, and 1.41 eV based on absorption onsets. Frontier orbital energy levels of ZITI-X are determined by cyclic voltammetry (see Figure S2), and the energy diagram including J71 donor is presented in Figure 2B. The HOMO and LUMO energy levels are calculated from onsets of oxidation and reduction processes that are calibrated by ferrocene/ferrocenium. ZITI-S, ZITI-C, and ZITI-N show slightly different LUMOs of −3.87, −3.81, and −3.82 eV, but significantly up-shifted HOMOs of −5.80, −5.65 to −5.53 eV, respectively. Thus, the HOMO offsets between ZITI-X and J71 (HOMO: −5.48 eV) gradually decrease for S, C, N molecules, suggesting reduced hole side driving force. As shown in Figure S3, the J71:ZITI-S, J71:ZITI-C, and J71:ZITI-N blends show high photoluminescence quenching efficiency indicating efficient charge separation.

Figure 2

PV Performance of J71:ZITI-X-Based OSCs

(A) Normalized UV-vis-NIR absorption spectra of ZITI-X and J71 in thin films.

(B) Energy diagram of donor and acceptor materials.

(C) Characteristic J-V curves.

(D) The corresponding EQE curves of J71:ZITI-X-based devices.

(A) Normalized UV-vis-NIR absorption spectra of n class="Chemical">ZITI-X and J71 in thin films.

Photovoltaic Characterization of ZITI-X-Based Devices

The conventional device architecture of ITO/PEDOT:PSS/active layer/PDINO/Al is adopted to evaluate photovoltaic performance of these acceptors. We screened the donor-acceptor (D-A) weight ratio and the temperature for thermal annealing treatment. Detailed photovoltaic parameters are summarized in Tables S1–S3. We also try other optimized approaches, such as solvent vapor annealing and adding additives, and the Detailed photovoltaic parameters are shown in Table S9. We find that a D-A weight ratio of 1:1 in chloroform and thermal annealing at 120°C for ZITI-C and ZITI-N and 100°C for ZITI-S are the best device fabrication conditions. As shown in Figure 2C and Table 1, J71:ZITI-S-based devices show relatively low PCEs of 8.86 ± 0.18% with a Voc of 0.812 ± 0.004 V, a Jsc of 17.18 ± 0.38 mA cm−2, and fill factor (FF) of 63.51 ± 0.87%; J71:ZITI-C and J71:ZITI-N-based devices exhibit high PCEs of 13.02 ± 0.13% and 13.47 ± 0.12% with Vocs of 0.851 ± 0.006, 0.873 ± 0.005 V, Jscs of 21.28 ± 0.26, 21.73 ± 0.33 mA cm−2, and FF of 72.03 ± 0.79%, 70.96 ± 0.88%, respectively. The corresponding histogram of PCE counts for 25 individual devices is displayed in Figure S7. Also, ZITI-N possesses an excellent compatibility with commercially available polymer donors (See Table S11). Despite the lowest optical bandgap of ZITI-N, the J71:ZITI-N blend gives the highest Voc with the lowest Eloss of 0.53 eV than J71:ZITI-S and J71:ZITI-C blends with Elosss of 0.80 and 0.62 eV, which can be correlated to HOMO energy levels. The Jscs are confirmed by EQE measurements (see Figure 2D). ZITI-N-based devices show a slightly lower EQEmax of 72.50% than ZITI-S- and ZITI-C-based ones with EQEmax of 74.35% and 78.01% in the NIR region, suggesting that the driving force (HOMO energy offsets) corresponds to carrier extraction. Such observation inspires us to blend ZITI-C and ZITI-N together, to balance the driving force and absorption (Liu et al., 2018b, Zhang et al., 2018c). We fabricated the ternary OSCs and optimized the acceptor ratios and thickness (See Tables S4 and S7). ZITI-C/N (1:1) HOMO energy level can be tuned to −5.60 eV (see Figure 2B), leading to a slightly increased HOMO offset (hole transfer driving force). As expected, a PCEavg of 14.05 ± 0.21% with a Voc of 0.857 ± 0.004 V, a Jsc of 23.01 ± 0.24 mA cm−2, and an FF of 71.72 ± 0.98% is obtained (see Figure 2C). The increased Jsc accounts for the improved PCE. Accordingly, a higher NIR EQEmax of 77.07% is obtained, suggesting improved charge separation. Moreover, the thick-film device is very important to meet the needs of future roll-to-roll mass production and lay the foundation for the commercial applications. We tried to fabricate the J71:ZITI-C:ZITI-N-based ternary devices with the different thickness (Table S12). The PCE still maintains over 12% when the thickness of the active layer increases to 200 nm.

Table 1

Photovoltaic Parameters of J71:ZITI-X-Based Devices

Acceptors	Voc (V)	bJsc (mA cm−2)	FF (%)	PCE (%)	Egopt (eV)	Eloss (eV)
ZITI-S	0.811 (0.812 ± 0.004)	17.39 (17.18 ± 0.38)	64.62 (63.51 ± 0.87)	9.12 (8.86 ± 0.18)	1.61	0.80
ZITI-C	0.851 (0.851 ± 0.006)	21.30 (21.28 ± 0.26)	72.76 (72.03 ± 0.79)	13.18 (13.02 ± 0.13)	1.47	0.62
ZITI-N	0.876 (0.873 ± 0.005)	21.78 (21.73 ± 0.33)	72.00 (70.96 ± 0.88)	13.68 (13.47 ± 0.12)	1.41	0.53
ZITI-C:ZITI-N	0.859 (0.857 ± 0.004)	23.05 (23.01 ± 0.24)	72.51 (71.72 ± 0.98)	14.36 (14.05 ± 0.21)	1.41	0.55

Average values with standard deviation were obtained from 25 devices.

To investigate the charge transport property in the blend films, the elen class="Chemical">ctron and hole mobilities are measured by using the space-charge-limited-current method (see Figure S4 and Table S5). The J71:ZITI-S device shows electron and hole mobilities of 1.54 × 10−4 and 2.75 × 10−4 cm2 V−1 s−1. J71:ZITI-C, J71:ZITI-N, and J71:ZITI-C:ZITI-N devices show the electron/hole mobilities of 1.86 × 10−4/3.83 × 10−4, 2.12 × 10−4/2.69 × 10−4, and 2.24 × 10−4/2.82 × 10−4 cm2 V−1 s−1, respectively. To investigate the carrier recombination under the short-circuit condition, we measured the current density at different light intensities (see Figure S9). All devices displayed the same relationship of Jsc = P0.95, indicating the weak bimolecular recombination for these blends. The dependence of Voc on the light intensity was measured resulting in slopes of 1.23, 1.32, 1.15, and 1.18 kT/q for J71:ZITI-S, J71:ZITI-C, J71:ZITI-N, and J71:ZITI-C:ZITI-N devices, respectively, implying low trap-assisted recombination.

Thin-film morphology of neat materials and BHJ blends were investigated using grazing incidence wide-angle X-ray diffraction (GIWAXS) and resonant soft X-ray scattering (RSoXS) (see Figure 3). J71 shows a face-on orientation with (100) at 0.29 Å−1 and a π-π stacking at 1.69 Å−1. ZITI-X series in neat films show similar molecular stacking of face-on orientations. The (100) peak is located at ∼0.32 Å−1 and the π-π stacking peak is located at 1.72, 1.85, and 1.70 Å−1 for ZITI-S, ZITI-C, and ZITI-N, respectively. Thus, detailed change in bridge heteroatoms can affect the π-π stacking of conjugated backbone. BHJ thin films show similar diffraction features for all these samples, except slight changes in the π-π stacking position, containing both J71 and ZITI-X structure information. RSoXS at 285.2 eV photon energy was used to investigate the phase separation of BHJ thin films. J71:ZITI-S shows a scattering hump at 0.014 Å−1, indicating a phase separation of 45 nm. J71:ZITI-C and J71:ZITI-N show quite similar scattering feature, with a scattering hump around 0.01 Å−1, indicating a phase separation of 63 nm. J71:ZITI-C:ZITI-N ternary blends retains a similar phase separation. Such morphology information is in good agreement with atomic force microscopy (AFM) and transmission electron microscopy (TEM) characterizations. The AFM under typing mode and TEM are utilized to detect the morphologies of blend films. As shown in Figure S6, J71:ZITI-S, J71:ZITI-C, J71:ZITI-N, and J71:ZITI-C:ZITI-N blends display very uniform and smooth surface with root-mean-square (RMS) roughnesses of 0.73, 0.86, 1.19, and 0.94 nm, respectively. TEM images are quite consistent with AFM in the changes of the fibrillar structures and self-aggregation of the blend films.

Figure 3

Morphology Investigations

(A) GIWAXS 2D patterns for ZITI-X pristine films and J71:ZITI-X blend films.

(B and C) The corresponding line-cuts of GIWAXS patterns of neat films and BHJ films.

(D) RSoXS profiles for J71:ZITI-X blend films.

(A) GIWAXS 2D patterns for ZITI-X pristine films and J71:n class="Chemical">ZITI-X blend films.

Energy Loss Analysis

To understand the origin of Voc difference, study on the energy loss pathways in J71:n class="Chemical">ZITI-X blends were conducted with Fourier-transform photocurrent spectroscopy, external quantum efficiency (FTPS-EQE), electroluminescence (EL), and electroluminescence quantum efficiency (EQEEL) measurements. The total energy loss Eloss can be attributed to two parts: energy loss due to charge generation (ΔE2 = Eg-ECT) and energy loss due to charge recombination (ECT-qVoc) that can be divided into the radiative (ΔE1) and non-radiative (ΔE3) recombinations.4,5 The ECT of each blend systems can be calculated by fitting the sub-gap absorption of the corresponding FTPS-EQE curves using the following equation (Vandewal et al., 2010):where EQEPV(E) is the photovoltaic external quantum efficiency, k is the Boltzmann constant, T is the absolute temperature (300 K in this work), and λ is the reorganization energy and f can be viewed as the interaction strength between donor and acceptor in the blends. Eloss can be calculated according to the following equation (Qian et al., 2018, Vandewal et al., 2010):where h is the Plank constant, q is the elementary charge, and c is the speed of light.

Figure 4A shows the normalized EL spen class="Chemical">ctra of the devices based on pure materials and blends. The J71:ZITI-S blend shows a single EL emission peak at 1,010 nm, which is assigned to the emission from the sub-gap charge transfer (CT) state. The strong CT state density can be a result of large energy level offsets (HOMO and LUMO); thus, the charge generation energy loss is strong. For the J71:ZITI-C blend, a weak and new CT emission peak also appears at around 1,010 nm together with another main emission band at around 906 nm assigned to ZITI-C; thus, such blend has a low CT state density that can be quickly filled and then to generate ZITI-C emission. The J71:ZITI-N blend only shows EL from ZITI-N, without noticeable CT states emission. Thus, quite low charge generation loss is presented owing to the small energy level offset. The FTPS-EQE curves (see Figure 4B) are used to obtain CT energy through the above-mentioned equation. The CT energy for J71:ZITI-S blend is 1.36 eV, resulting in an Eg-ECT energy offset (ΔE2: 0.30 eV). In contrast, both J71:ZITI-C and J71:ZITI-N blends show higher ECT (1.45 V for J71:ZITI-C and 1.44 eV for J71:ZITI-N) leading to much smaller Eg-ECT offsets (ΔE2: 0.10 eV for J71:ZITI-C and 0.07 eV for J71:ZITI-N). Because the Voc of OSCs is directly restricted by the energy of the CT state, J71:ZITI-C and J71:ZITI-N blends with high CT energies could achieve higher Voc compared with that of the J71:ZITI-S blend. The energy losses due to radiative recombination (ΔE1) of charge carriers are calculated using the fitted parameters from the FTPS-EQE spectra, and the losses due to the non-radiative recombination (ΔE3) can be quantified by measuring the EQEEL. As ΔE1 is unavoidable for all kinds of solar cells, it is important to maximize EQEEL to minimize ΔE3. As shown in Figure 4C and Table S6, the emission efficiency of J71:ZITI-N blend is 8.16 × 10−4%, one order of magnitude higher than those of the other two blends, 2.54 × 10−5% (J71:ZITI-S) and 4.39×10−5% (J71:ZITI-C), which represents significantly decreased non-radiative recombination losses (0.30 eV) compared with other blends (0.39 eV for J71:ZITI-S and 0.38 eV for J71:ZITI-C). Therefore, we attribute the highest Voc of J71:ZITI-N blend to the smallest Eg-ECT energy offset and the suppressed non-radiative recombination losses (Figure 4D). For the driving force that leads to the charge generation (ΔE2 = Eg-ECT) or in other words the energy loss generated during the exciton dissociation process, the J71:ZITI-S blend has the largest value of 0.30 eV due to large energy offsets in both HOMO and LUMO. ZITI-C and ZITI-N have quite similar LUMO energy levels, and thus the difference in driving force comes from the energy offsets from HOMO side. Thus, manipulating the HOMO energy level is an important route in mitigating the energy loss in BHJ blends. The energy loss analysis of ternary OSCs are shown in Figure S5 and Table S6. In the ternary blends, whose chemical similarity makes molecular alloy possible, make a useful platform to fine-tune frontier energy levels. Such a strategy can effectively manipulate the detailed energy loss and driving force, two factors that interactively affect Voc and Jsc. Such an optimization yields improved device performances with a maximized PCE of 14.36%, a success of physical understanding in improving efficiency of using existing materials.

Figure 4

Energy Loss Analysis

(A) Electroluminescence spectra of devices based on the pristine NFAs and blended films.

(B and C) FTPS-EQE and EQEEL of the blended devices.

(D) The comparison of ΔE1, ΔE2, and ΔE3 of J71:ZITI-X-based devices.

Hole-Transfer Kinetics and Driving Force Correlation

To quantitatively assess the HOMO energy offset, or driving force on photoinduced hole transfer process, we performed femtosecond (fs) transient absorption (TA) spen class="Chemical">ctroscopy to directly probe the photo-induced hole transfer dynamics in J71:ZITI-S, J71:ZITI-C, J71:ZITI-N, and J71:ZITI-C:ZITI-N blend films (see Figure 5). The steady-state absorption peaks for different acceptors and donors are well separated in spectral domain; therefore, both spectral and temporal characteristics of charge transfer dynamics can be extracted. To investigate hole transfer, the excitation wavelength of 750 nm was selected here to excite only acceptors. The color plot of fs TA spectra of J71:ZITI-S film after 750-nm excitation are shown in Figure 5A and a few representative TA spectra at indicated delay times in Figure 5B with TA spectrum of neat ZITI-S film (gray circles) also shown for comparison. The bleach peaks at ∼650 and 730 nm appear in both neat ZITI-S and J71:ZITI-S films, corresponding to the ground state bleach and the stimulated emission of the lowest energy transition (S1) in ZITI-S due to photoexcitation. With the decay of ZITI-S bleach peak at 630–760 nm, a few new bleach peaks at 550–620 nm appear in the TA spectrum of J71:ZITI-S film. These peaks at 550–620 nm match very well with the absorption features of J71 films. The bleach decay process of photoexcited ZITI-S (630–760 nm) agrees well with the rise process of the J71 ground state bleach (550–620 nm), confirming photoexcited hole transfer from ZITI-S to J71. The hole transfer in J71:ZITI-C, J71:ZITI-N, and J71:ZITI-C:ZITI-N blend films are also observed as shown in Figure S8 in spite of smaller HOMO energy offsets. The ground state bleach of J71 in J71:ZITI-S, J71:ZITI-C, and J71:ZITI-N blend films rise with a half-time of ∼3.2, 4.7, and 20 ps, respectively, corresponding to the hole transfer lifetime. This trend of hole transfer rate agrees with the value of HOMO offsets. Thus, a higher driving force is critical in ensuring efficient carrier extraction from NFA acceptors to enhance Jsc. The high Jsc obtained from J71:ZITI-N blends is attributed to the extended absorption. Blending ZITI-C with ZITI-N recovers the hole-transfer lifetime to 7.0 ps. Such an operation better balances the light absorption and extraction efficiency, yielding the highest Jsc among BHJ devices. We also investigated the photo-induced electron transfer process by exciting the J71 donor and probing its excited state absorption at 1,175 nm. The decay kinetics of the J71 excited state in different blend films are shown in Figure 5D, which shows that electron transfers in all blend films are very efficient and are shorter than 1 ps. Thus, hole-transfer process is the major barrier in generating device current.

Figure 5

Transient Absorption Analysis

(A) Color plot of fs Transient absorption spectra of J71:ZITI-S blend film under 750 nm excitation with a fluence below 10 μJ/cm2.

(B) Representative fs Transient absorption spectra of J71:ZITI-S blend film at indicated delay times; Gray dots: TA spectrum of neat ZITI-S film at 1 ps excited by 750 nm.

(C) TA kinetics of J71:ZITI-S, J71:ZITI-C, J71:ZITI-N, and J71:ZITI-C:ZITI-N blend films showing hole transfer process.

(D) TA kinetics of excited state absorption (1175 nm) of J71:ZITI-S, J71:ZITI-C, J71:ZITI-N, and J71:ZITI-C:ZITI-N blend films showing electron transfer process.

(A) Color plot of fs Transient absorption spectra of J71:n class="Chemical">ZITI-S blend film under 750 nm excitation with a fluence below 10 μJ/cm2.

(B) Representative fs Transient absorption spectra of J71:n class="Chemical">ZITI-S blend film at indicated delay times; Gray dots: TA spectrum of neat ZITI-S film at 1 ps excited by 750 nm.

(C) TA kinetics of J71:ZITI-S, J71:n class="Chemical">ZITI-C, J71:ZITI-N, and J71:ZITI-C:ZITI-N blend films showing hole transfer process.

(D) TA kinetics of excited state absorption (1175 nm) of J71:ZITI-S, J71:n class="Chemical">ZITI-C, J71:ZITI-N, and J71:ZITI-C:ZITI-N blend films showing electron transfer process.

Driving Force, Energy Loss, Charge Extraction, and Absorption Manipulation

The previous discussions reveal the importance of manipulating frontier energy levels to better balance driving force, n class="Disease">energy loss, and charge extraction. We make J71:ZITI-C:ZITI-N ternary blends of more blending ratios to establish a correlation. Thin film absorption and internal quantum efficiency (IQE) were first explored. As shown in Figure 6A, the J71:ZITI-N blend film shows the stronger absorbance of 0.654 at the maximum absorption of 793 nm than J71:ZITI-C, 0.466 at 750 nm. However, as shown in IQE curves (see Figure 6B), the J71:ZITI-N blend delivers a weaker IQE response of 79.47% (793 nm) compared with that of the J71:ZITI-C blend, 83.05% (750 nm). Thus, despite the largest spectral overlap with sunlight of ZITI-N that is essential to get a large Jsc, the J71:ZITI-N blend gives only a comparable Jsc of 21.73 ± 0.33 mA cm−2 as that of ZITI-C, 21.28 ± 0.26 mA cm−2. Such result clearly demonstrates the complex interconnection between material bandgap and system driving force in generating current. A simple 1:1 mixing, although low in peak absorption (790 nm), elevates the driving force of hole transfer, giving rise to a high IQE value of 86.05%. And thus a higher Jsc of 23.01 ± 0.24 mA cm−2 is obtained. A more detailed correlation can be seen in ternary blends of different compositions. As expected, high ZITI-N loading could reduce the effective HOMO energy level in the mixture, and thus a low energy loss and a high Voc can be obtained, as seen from the monotonic increase in Figure 6C. In Jsc spectrum, a higher ZITI-N loading should linearly increase Jsc if only absorption is considered. However, high Jsc is seen in mixtures, with the highest value at 1:1 ratio. Such result unequivocally reveals the importance of driving force in helping charge extraction, ensuring high device Jsc. And thus the maximum efficiency is obtained when ZITI-C:ZITI-N = 1:1 (Figure 6D). Thus, the detailed balance among driving force, energy loss, charge extraction, and light absorption should be carefully manipulated in high efficiency OPV devices. Interestingly, by changing the donor J71 to PBDB-TF, the ternary OSCs-based PBDB-TF:ZITI-C:ZITI-N achieve a higher PCE of 13.85% with simultaneously elevated Jsc and FF than the binary OSCs (Table S10).

Figure 6

Current and Voltage Balance Analysis

(A and B) BHJ blends absorption profiles and internal quantum efficiency of J71:ZITI-C:ZITI-N binary and ternary blends.

(C and D) Voc and Jsc dependence on ZITI-N ratio in ternary blends.

(A and B) BHJ blends absorption profiles and internal quantum efficiency of J71:n class="Chemical">ZITI-C:ZITI-N binary and ternary blends.

Conclusions

To conclude, we achieved a panoramic understanding of the relationship between driving force and OSCs performance with a new family of NIR electron acceptors, ZITI-S, ZITI-C, and ZITI-N, possessing high structural similarity. Among the three acceptors, ZITI-N exhibits the highest HOMO energy level and the lowest optical bandgap of 1.41 eV with an absorption onset at 879 nm. When blended with the wide-bandgap polymer donor, J71:ZITI-N-based devices deliver the highest PCEavg of 13.47 ± 0.12% at the smallest Eloss of 0.53 eV than those of J71:ZITI-S- and J71:ZITI-C-based devices with PCEavgs of 8.86 ± 0.18% and 13.02 ± 0.13% and Elosss of 0.80 and 0.62 eV, respectively. Furthermore, by combining ZITI-C and ZITI-N acceptors, a higher PCEavg of 14.05 ± 0.21% with an improved Jsc of 23.01 ± 0.24 mA cm−2 is achieved, which can be ascribed to the slightly increased driving force according to the detailed study on the energy loss channels. The fs-TA spectroscopy investigation indicates that the J71:ZITI-N blend with the lowest HOMO offset shows the slowest hole transfer rate. We observed that the CT emissions gradually decreased and finally disappeared from J71:ZITI-S, J71:ZITI-C to J71:ZITI-N blends, which means that, when the active layers have both optimized morphology and excellent absorption, by narrowing HOMO offset between donor and acceptor materials, the superiority of NFAs can be fully realized. Investigations by implementing PL quenching and EL experiments provide valuable information on the efficiency of charge separation and the magnitude of driving force, which may accelerate the donor:acceptor screening and relieve the workload from the tedious device fabrication. Moreover, with the ideal research models based on J71 and ZITI-X, we unambiguously reveal that a meticulous control of HOMO offset is critical to approaching the efficiency limit in NFA OSCs.

Limitations of Study

In our research system, when we investigate the relationship between HOMO offset and the efficiency for charge separation, we can exclude the impan class="Chemical">ct originating from the blend morphology because of the high structural similarity of ZITI-X NIR acceptors. However, the morphology impact should be carefully considered in other systems.

Methods

All methods can be found in the accompanying Transparent Methods supplemental file.