Introduction

Solution-processable organic photovoltaics (OPVs) constitute an important next-generation photovoltaic technology featuring flexibility, light weight, and semitransparency [1,2,3]. Recent progress in the use of π-conjugated materials and polymers as donors and acceptors, particularly small-bandgap nonfullerene acceptors (NFAs), has dramatically improved power conversion efficiencies (PCEs) [4,5,6]. For π-conjugated polymer donors, a promising molecular design strategy involves extension of the π-electron system of the building unit [7, 8]. For example, most of the high-performance polymers reported recently incorporate benzo[1,2-b:4,5-b’]dithiophene (BDT), a relatively electron-rich building unit, in their polymer backbones [9,10,11]. Additionally, π-extensions of electron-deficient heteroaromatic cobuilding units have also been studied [12]. Among a number of electron-deficient heteroaromatic systems, thiazole-based fused rings such as thiazolo[5,4-d]thiazole (TzTz) and benzo[1,2-d:4,5-d’]bisthiazole (BBTz) have been shown to be excellent systems for building π-conjugated polymers (Fig. 1a, b) [13,14,15]. Indeed, π-conjugated polymers consisting of BDT and alkylthiophene-flanked TzTz or BBTz showed relatively high PCEs in NFA-based cells [16,17,18,19,20,21,22].

Fig. 1
figure 1

Chemical structures of (a) thiophene-TzTz-thiophene and (b) thiophene-BBTz-thiophene building units with alkyl, alkoxy, and ester groups, and (c) BBTz-based polymers, PDBTz1 and PDBTz2

Full size image

The use of noncovalent intramolecular interactions to enable coplanarity of the polymer backbone is also an important molecular design strategy for the development of high-performance π-conjugated polymers [23]. The fluorine atom is a common substituent that interacts with hydrogen and sulfur atoms; thus, fluorine is often introduced on heteroaromatic rings linked with thiophenes or thiazoles [24,25,26]. The oxygen atom can also interact with the sulfur atom; thus, alkoxy and ester groups are often used as side chains instead of alkyl groups on heteroaromatic rings linked with thiophene or thiazole [27, 28]. For example, our group reported the use of alkoxythiophene- and esterthiophene-flanked TzTz as building blocks for π-conjugated polymers, which were more crystalline than their alkyl counterparts (Fig. 1a) [29, 30]. As a lower highest occupied molecular orbital (HOMO) energy level is preferred for donor materials to provide high open-circuit voltage (VOC) in OPV [31], electron-withdrawing ester groups are typically chosen for the side chains rather than electron-donating alkoxy groups. However, for polymers with inherent very low HOMO energy levels, the alkoxy group functions better than the ester group. Herein, we synthesized alkoxythiophene-flanked BBTz as a novel building unit (Fig. 1b) and two simple BBTz-based polymers incorporating bithiophene counits, i.e., PDBTz1, which has alkyl groups on all of the bithiophene units, and PDBTz2, which has alternating alkyl and alkoxy groups on the bithiophene units (Fig. 1c). Notably, when the polymers were blended with Y6, a benchmark NFA, hole transfer was inefficient in the PDBTz1:Y6 blend because the offset energy of the HOMOs was too small, hole transfer was significantly improved in the PDBTz2:Y6 blend due to the sufficiently large offset originating from the elevated HOMO energy level in PDBTz2. Interestingly, whereas PDBTz1 exhibited a highly crystalline structure with nearly complete edge-on orientations, PDBTz2 exhibited a highly crystalline structure with nearly complete face-on orientations, which resulted in greater out-of-plane charge transport in PDBTz2 than in PDBTz1. As a result, PDBTz2 showed greater photovoltaic performance than PDBTz1.

Syntheses

The synthetic routes to the BBTz-bithiophene copolymers are shown in Scheme 1. The ring-closing reaction of 2,5-diaminobenzene-1,4-dithiol dihydrochloride (1) and 3-(2-hexyldecyl)thiophene-2-aldehyde (2) afforded an alkylthiophene-flanked BBTz derivative (3) in 35% yield [32]. Subsequently, 3 was dibrominated with N-bromosuccinimide (NBS) in a chloroform (CF)/acetic acid (AcOH) cosolvent system to give the dibrominated BBTz monomer (4) in 80% yield. 4 was then distannylated via lithiation with n-butyl lithium and subsequent treatment with trimethyltin chloride to provide 5 in 76% yield. An alkoxythiophene-flanked BBTz derivative (7) was obtained in 15% yield via a ring-closing reaction using 1 and 3-(2-hexyldecyloxy)thiophene-2-carbaldehyde (6) and then was dibrominated to give 8 in 75% yield. 4 was copolymerized with 5 and 8 to afford PDBTz1 and PDBTz2, respectively, via the Stille coupling reaction with Pd2(dba)3·CHCl3 and P(o-tol)3 as the catalyst system. The molecular weights of the polymers were determined with high-temperature gel permeation chromatography (GPC) with trichlorobenzene (TCB) as the eluent at 180 °C. The number-average molecular weight (Mn) and dispersity (Đ) were 36,200 and 1.9 for PDBTz1 and 33,500 and 2.3 for PDBTz2, respectively (Fig. S1). Interestingly, both polymers were soluble in CF and CB even at room temperature, even though BBTz had a π-extended building unit without substituents. Differential scanning calorimetry showed no phase transitions below 350 °C (Fig. S2), indicating that both polymers had good thermal stabilities.

Scheme 1
scheme 1

Synthetic route to BBTz-based polymers with alkyl groups (PDBTz1) and alkoxy groups (PDBTz2). HD = 2-hexyldecyl

Full size image

Polymer properties

Cyclic voltammetry (CV) was used to evaluate the energy levels of the polymers. Figure 2a shows the cyclic voltammograms of the polymers. The HOMO and LUMO energy levels (EHOMO and ELUMO, respectively) were determined from the onset redox potentials, which were −1.99 and 0.84 V for PDBTz1 and −1.97 and 0.67 V for PDBTz2, respectively, after calibration with ferrocene. The energy levels of the polymers, along with those of Y6, a benchmark NFA, are summarized in Fig. 2b and Table 1. The EHOMO of PDBTz2 was −5.47 eV, which was 0.17 eV greater than that of PDBTz1 (−5.64 eV). In addition, photoemission yield spectroscopy (PYS) indicated that PDBTz2 had a higher EHOMO (−5.01 eV) than PDBTz1 (−5.28 eV) by 0.27 eV (Fig. S3). The higher HOMO energy level of PDBTz2 was explained by the stronger electron-donating nature of the alkoxy group relative to that of the alkyl group. As a result, the HOMO energy offset between the polymer and Y6 was 0.22 eV for PDBTz2, which was larger than that for PDBTz1 (0.05 eV). On the other hand, both polymers had similar LUMO energy levels of approximately −2.80 eV, which was likely dependent on the electron-deficient BBTz unit. Thus, PDBTz2 is expected to have a smaller bandgap than PDBTz1. These trends were roughly consistent with the results from DFT calculations at the B3LYP 6-31G(d) level (Fig. S4).

Fig. 2
figure 2

a Cyclic voltammograms of the polymers in the thin films. b Energy diagrams of the polymers along with Y6. The value in brackets is the EHOMO determined by the PYS measurement. c UV–vis absorption spectra of the polymers in thin films. Temperature-dependent UV–vis absorption spectra of (d) PDBTz1 and (e) PDBTz2 in CB solution. f Energy variations of the thiazole–methylthiophene (Tz–MeT) and thiazole–methoxythiophene (Tz–MeOT) linkages as a function of dihedral angle calculated with the DFT method (B3LYP/6-31G(d) level)

Full size image
Table 1 Optical properties of the polymers
Full size table

The UV–vis spectra of the polymer thin films are shown in Fig. 2c, and the corresponding optical parameters are listed in Table 1. PDBTz1 exhibited two absorption maxima (λmax) at 557 and 515 nm, which correspond to the 0–0 and 0–1 transitions, respectively. PDBTz2 also had λmax values of 580 and 536 nm, which were ~20 nm greater than those of PDBTz1. The optical bandgap (Egopt) for PDBTz2 calculated with the absorption onset (λonset) was 2.01 eV (λonset = 616 nm), which was smaller than that for PDBTz1 (Egopt = 2.11 eV, λonset = 589 nm). Interestingly, the intensity ratio of the 0–0 band with respect to the 0–1 band (I0–0/I0–1) was greater in PDBTz2 (1.13) than in PDBTz1 (0.97). We also noted that PDBTz2 had a significantly greater absorption coefficient than PDBTz1 in both the solution and thin films (Fig. S5 and Table S2), which was consistent with the computations (Fig. S6). These results implied that the polymer backbone in PDBTz2 was more coplanar and rigid than that in PDBTz1 [33, 34].

We also measured the temperature-dependent UV–vis absorption spectra of the polymers in the CB solution (Fig. 2d, e). However, in PDBTz1, the intensity of the 0–0 band was significantly lower than that of the 0–1 band at 100 °C, whereas in PDBTz2, the intensity of the 0–0 band was still similar to that of the 0–1 band. This suggested that PDBTz2 has a more rigid backbone and greater aggregation than PDBTz1. This was validated by plotting the energy as a function of the dihedral angle for the thiazole–methylthiophene (Tz–MeT) and thiazole–methoxythiophene (Tz-MeOT) linkages, as models for PDBTz1 and PDBTz2; these were calculated with the DFT method (B3LYP/6-31G(d) level) (Fig. 2f). The torsional energy barrier, the difference between the energy at a dihedral angle of 90° and that at 0°, for Tz–MeOT was 8.3 kcal mol−1, which was ~1.5 times greater than that for Tz–MeT (5.4 kcal mol−1). This suggested that Tz–MeOT suppressed torsion more than Tz–MeT, which was due to the S···O noncovalent intramolecular interactions between the sulfur atom in the thiazole ring and the oxygen in the alkoxy group. Therefore, it is expected that the backbone of PDBTz2 is more coplanar than that of PDBTz1. Moreover, Tz–MeOT adopts the anti-conformation (a dihedral angle of 0°) rather than the syn-conformation (a dihedral angle of 180°) more readily than Tz–MeT, as the energy difference between these conformations was 4.8 kcal mol−1 for the former and 0.6 kcal mol−1 for the latter, which was likely due to the N···O Coulombic repulsion between the lone pairs on nitrogen in the thiazole and the methoxy oxygen. Therefore, PDBTz2 should predominantly exhibit an anti-conformation and provide a more regular backbone structure than PDBTz1, which would increase the backbone order and crystallinity.

Fig. 3
figure 3

a J–V curves and (b) EQE spectra of the polymer:Y6 cells. c Photoluminescence spectra of the Y6 neat film and polymer:Y6 blended films excited at 680 nm (Y6 excitation). c Photoluminescence spectra for a Y6 neat film and polymer:Y6 blend films excited at 680 nm (Y6 excitation)

Full size image

Photovoltaic characteristics

We fabricated OPV cells with conventional structures (ITO/PEDOT:PSS/polymer:Y6/PNDIT-F3N-Br/Ag). A polymer:Y6 blend with a weight ratio of 1:1.2 was spin-coated with a 6.0 g L−1 CF solution containing 0.5% (v/v) 1-chloronaphthalene. Figures 3a, b show the current density–voltage (J–V) curves and external quantum efficiency (EQE) spectra of the cells, and the photovoltaic parameters are listed in Table 2. The PDBTz2 cell showed a short-circuit current density (JSC) of 26.0 mA cm−2, which was approximately twice that of the PDBTz1 cell (12.8 mA cm−2). Correspondingly, the maximum EQE of the PDBTz2 cell (85%) was also approximately twice that of the PDBTz1 cell (42%). The JSC value calculated from the EQE spectra (JSCEQE) was consistent with the JSC value obtained from the J–V measurements (Table 2). The PDBTz2 cell exhibited a lower VOC (0.73 V) than the PDBTz1 cell (0.77 V), which was consistent with the higher EHOMO of PDBTz2 relative to that of PDBTz1. The fill factors (FF) were relatively low for both cells: it was 0.47 for PDBTz2 and 0.49 for PDBTz1. As a result, the PDBTz2 cell showed a PCE of 8.8%, which was much greater than that of the PDBTz1 cell (4.6%).

Table 2 Photovoltaic properties of the polymer:Y6 cells
Full size table

We also tested OPV cells with inverted structures (ITO/ZnO/polymer:Y6/MoOx/Ag), in which the blended films were fabricated similar to those of conventional cells. For PDBTz2, the inverted cell provided a lower JSC (22.7 mA cm−2) but a higher VOC (0.79 V) and FF (0.61) than the conventional cells, leading to higher PCEs of up to 10.9%. This PCE was also higher than that of the inverted cell for PDBTz1 (PCE = 3.8%, JSC = 9.4 mA cm−2, VOC = 0.80 V, FF = 0.50).

To understand the significant difference in the JSC values for the PDBTz1 cell and the PDBTz2 cell, we investigated the photoluminescence (PL) quenching efficiencies of the blended films (Fig. 3c). When excited at 560 nm (polymer excitation), the quenching efficiencies were nearly 100% for both the PDBTz1:Y6 and PDBTz2:Y6 blended films. Quenching of the polymer PL was likely dominated by energy transfer (Fig. S7). In contrast, when excited at 680 nm (Y6 excitation), the quenching efficiency was 12% for the PDBTz1:Y6 blend film and 92% for the PDBTz2:Y6 blend film. This indicated that, whereas hole transfer from Y6 to PDBTz1 was inefficient, hole transfer from Y6 to PDBTz2 was highly efficient. Because there were no significant differences in the morphologies of the two films (Fig. S8), the inefficient hole transfer of the PDBTz1:Y6 blended film was ascribed to the small energy offset between the EHOMOs, which accounted for the low JSC in the PDBTz1 cell.

Packing orders of the polymers

Grazing-incidence X-ray diffraction (GIXD) measurements were performed to investigate the packing orders of the polymers. Figure 4a–d depicts the two-dimensional (2D) GIXD patterns and cross-sectional diffraction profiles (Fig. 4e, f) of the neat and blended polymer films. The diffraction parameters of the neat and blended polymer films are listed in Table 3. In the neat films (Fig. 4a, b), PDBTz1 adopted an edge-on orientation, as the (0 1 0) π–π stacking diffraction (qxy ≈ 1.70 Å−1) and (h 0 0) lamellar diffractions (~qz ≈ 0.33 Å−1) appeared along the qxy and ~qz axes, respectively. Moreover, PDBTz2 adopted a face-on orientation, as the (0 1 0) π–π stacking diffraction peak (~qz ≈ 1.75 Å−1) and (h 0 0) lamellar diffraction peak (qxy ≈ 0.29 Å−1) appeared along the ~qz and qxy axes, respectively (Fig. 4e). The π–π stacking distance (dπ) of PDBTz2 was 3.55 Å, which was significantly shorter than that of PDBTz1 (3.69 Å). Although the coherence length of the π–π stacking diffraction (CLπ) for PDBTz2 (31.9 Å) was slightly smaller than that for PDBTz1 (37.7 Å), the values should not be directly compared because these diffraction angles appeared in the wide-angle region of the different q axes. Both polymers exhibited higher-order lamellar diffractions, up to the fourth order, which indicated their highly crystalline natures. Notably, such higher-order lamellar diffractions are rarely observed for polymers with face-on orientations because highly crystalline polymers typically exhibit edge-on orientations [35]. The lamellar distance (dL) of PDBTz2 (21.3 Å) was greater than that of PDBTz1 (18.8 Å). This difference was attributed to the 2-hexyldecyloxy group in PDBTz2, which were longer than the 2-hexyldecyl groups in PDBTz1. The coherence length corresponding to the (1 0 0) lamellar structure (CLL) was 164 Å for PDBTz2, which was significantly greater than that of PDBTz1 (138 Å). Overall, we concluded that PDBTz2 was more crystalline than PDBTz1, which was attributed to the greater backbone coplanarity of PDBTz2.

Fig. 4
figure 4

a–d 2D GIXD patterns of the polymer neat films and polymer:Y6 blended films. a PDBTz1, b PDBTz2, c PDBTz1:Y6, and d PDBTz2:Y6. e, f Cross-sectional diffraction profiles cut from the 2D GIXD patterns along the qxy (solid line) and ~qz (dashed line) axes. e Polymer neat films. f Polymer:Y6 blended films

Full size image
Table 3 Structural parameters extracted from the GIXD measurement parameters of the polymer neat films and polymer:Y6 blend films
Full size table

The polymer:Y6 blend films exhibited clear diffraction along the ~qz axis corresponding to the (0 1 0) π–π stacking structures with a dπ of 3.51 Å (Fig. 4c, d, f). The CLπ for the PDBTz2:Y6 blended film was 31.4 Å, which was larger than that of the PDBTz1:Y6 blended film (26.0 Å). However, as the diffraction angles from the (0 1 0) π–π stacked structures of the polymer and Y6 should overlap, we could not fairly compare the crystallinities of the polymers in the blended films. Therefore, we instead used the polymer lamellar diffraction peak that appeared on the qxy axis. In fact, PDBTz2 showed greater CLL (146 Å) than PDBTz1 (51 Å), indicating that PDBTz2 had higher crystallinity than PDBTz1 even in the blended film.

Charge carrier transport

To investigate the in-plane charge carrier transport properties, we fabricated organic field-effect transistor (OFET) devices with top-gate/bottom-contact (TG/BC) architectures. The polymer layer was spin-coated from a CB solution and then annealed at 200 °C for 30 min. Figure 5a–c shows the transfer and output curves of the OFET devices, respectively, and Table 4 summarizes the OFET characteristics. Both polymers showed unipolar p-channel characteristics. Interestingly, although PDBTz2 had a face-on orientation, which is undesirable for in-plane charge carrier transport, the hole mobility (µhFET = 0.198 cm2 V−1 s−1) was comparable to that of PDBTz1 (0.203 cm2 V−1 s−1). This was probably due to enhanced intrachain charge carrier transport in PDBTz2, which originated from its highly coplanar and rigid backbone. In addition, the threshold voltage (Vth) of PDBTz2 (−24 V) was significantly lower than that of PDBTz1 (−37 V), which was most likely due to its higher-lying EHOMO, ensuring efficient hole injection from the electrode. The current on/off ratios (Ion/Ioff) were reasonably high for both polymers (1.4 × 105 for PDBTz1 and 1.9 × 105 for PDBTz2).

Fig. 5
figure 5

a Transfer and (b, c) output curves of OFET devices of the polymers

Full size image
Table 4 Charge carrier mobilities of polymers based on the FET
Full size table

We also investigated the out-of-plane charge carrier transport properties with the space-charge-limited current (SCLC) model (Fig. S9 and Table S3). Although the hole-only devices for the polymer neat film did not exhibit J–V curves that fit this model, the blended films provided good J–V curves that did fit. As a result, the PDBTz2:Y6 blended film exhibited a hole mobility (µhSCLC) of 1.1 × 10−4 cm2 V−1 s−1, which was significantly greater than that of the PDBTz1:Y6 blended film (3.9 × 10−5 cm2 V−1 s−1). This most likely occurred because the crystallinity of the blended film was greater for PDBTz2 than for PDBTz1, as described above. In addition, the electron mobilities (µeSCLC) were 6–7 × 10−5 cm2 V−1 s−1 for both blended films. The higher µhSCLC for the PDBTz2:Y6 blended relative to that for the PDBTz1:Y6 blended film may account for the higher JSC and FF in the PDBTz2 cell than in the PDBTz1 cell.

Conclusions

In this work, we synthesized a new building unit in which two alkoxythiophenes were attached to BBTz, which was copolymerized with an alkylthiophene-flanked BBTz building unit. The resulting polymer, named PDBTz2, showed a higher HOMO energy level as well as a narrower optical bandgap than its alkyl counterpart (PDBTz1) owing to the electronic effects of the electron-donating alkoxy groups. The higher HOMO energy of PDBTz2 led to a better matched energy offset of the HOMOs when blended with Y6, which facilitated charge transfer. Furthermore, with the S···O noncovalent intramolecular interactions between the alkoxy oxygens and the thiazole sulfurs in BBTz, PDBTz2 exhibited a more coplanar and rigid backbone and greater aggregation than PDBTz1, resulting in greater crystallinity. Interestingly, PDBTz2 exhibited a nearly complete face-on backbone orientation, while PDBTz1 exhibited an edge-on orientation. As a result, the PDBTz2:Y6 cells exhibited remarkably higher JSCs and thereby higher PCEs than the PDBTz1:Y6 cells. These results showed that alkoxythiophene-flanked BBTz has great potential for use as a building unit for π-conjugated polymers with highly crystalline and desirable face-on orientations. Further studies on polymers incorporating the building unit are currently underway in our group.