Highly efficient organic photovoltaics with enhanced stability through the formation of doping-induced stable interfaces

Significance

We have developed ultraflexible 3-µm-thick organic photovoltaics (OPVs) with good environmental stability and a high efficiency of over 13% by a simple postannealing process. By combining the stable polymer donor/nonfullerene small molecule acceptor blends and thermally stabilized interfaces due to doping, our OPVs achieve extremely long shelf lifetime with an estimated value of over 11.4 y. A laboratory-scale OPV module is fabricated with a sufficiently high efficiency of 11.4%, total area of 2.04 cm², and a maximum power per unit weight of 9.90 W/g, showing a promising future for flexible OPVs as power sources for wearable electronics and soft robotics.

Abstract

Flexible organic photovoltaics (OPVs) are promising power sources for wearable electronics. However, it is challenging to simultaneously achieve high efficiency as well as good stability under various stresses. Herein, we demonstrate the fabrication of highly efficient (efficiency, 13.2%) and stable OPVs based on nonfullerene blends by a single-step postannealing treatment. The device performance decreases dramatically after annealing at 90 °C and is fully recovered after annealing at 150 °C. Glass-encapsulated annealed OPVs show good environmental stability with 4.8% loss in efficiency after 4,736 h and an estimated T₈₀ lifetime (80% of the initial power conversion efficiency) of over 20,750 h in the dark under ambient condition and T₈₀ lifetime of 1,050 h at 85 °C and 30% relative humidity. This environmental stability is enabled by the synergetic effect of the stable morphology of donor/acceptor blends and thermally stabilized interfaces due to doping. Furthermore, the high efficiency and good stability are almost 100% retained in ultraflexible OPVs and minimodules which are mechanically robust and have long-term operation capability and thus are promising for future self-powered and wearable electronics.

Organic photovoltaics (OPVs) have emerged as promising renewable energy devices because of their advantages such as cost-effective fabrication over a large area, light weight, and flexibility/stretchability (1–4). Power conversion efficiency (PCE), which is one of the most important characteristics of OPVs, has been improved dramatically to 17.3% (5) and 16.5% (6) in tandem and single-junction OPVs on rigid glass and to 12.5% (7) on flexible polymer substrates.

For practical applications, apart from higher PCE, good environmental stability (including storage and thermal stability) is also important for the long-term stable operation of OPVs in harsh conditions. Degradation caused by oxygen, water, mechanical deformation, and light irradiation could be prevented or minimized by encapsulation (8), mechanical designing (9, 10), and ultraviolet (UV)-light filters (11). The remaining obstacle to the widespread application of OPVs is the achievement of long-term stability, which is very challenging (12, 13). For instance, the performance of OPVs can deteriorate dramatically during their initial operation, referred to as the “burn-in” loss (14). Highly efficient OPVs with fullerene acceptors tend to degrade significantly when heated at high temperatures, because of accelerated phase separation (15–17). Additionally, changes in the properties of the hole transport layer (HTL) might also cause device degradation. OPVs with a poly(3,4-ethylenedioxythiophene)–poly(styrenesulfonate) (PEDOT:PSS) HTL in conventional cells usually have limited lifetimes because PEDOT:PSS can react with the indium tin oxide (ITO) electrode and is sensitive to humidity (18), which makes it more challenging for flexible OPVs to achieve good stability due to unsatisfied barrier properties of thin polymer encapsulation layers. The chemical composition of metal oxide HTLs can change upon annealing, leading to dramatic device degradation (19).

Many strategies have been implemented, mainly on active layers, to improve the environmental stability of OPVs. They include polymer purification (14, 20), removal of additives (21), and the use of nonfullerene small molecules (22) or polymer acceptors (23) and ternary blends (24, 25). The state-of-the-art work on the storage stability shows that a ternary P3HT:IDTBR:IDFBR (1:0.7:0.3) device with a champion 7.7% PCE achieved a T₈₀ lifetime (defined as the time when a device loses 20% of its initial efficiency) of 1,200 h when stored unencapsulated in the dark with air at room temperature (RT) (25). With regard to the thermal stability, the state-of-the-art device based on a PBDBT:ITIC blend with an initial 11.2% PCE achieved a T₈₀ lifetime of >250 h after heating at 100 °C in a nitrogen-filled glove box (26). Apart from the active layers, the HTL can also influence the device stability. OPVs based on P3HT/PC₆₁BM and VO_x HTL with 3.9% PCE showed no degradation after annealing at 65 °C for 1,000 h, while OPVs based on V₂O₅ and PEDOT:PSS lost 60% and 13% of the initial efficiency under the same condition (27).

However, it is challenging to develop a simple approach without complicated processes to fabricate OPVs with >12% efficiency, good environmental stability, and areal scalability. Two difficulties must be overcome simultaneously to balance the trade-off between the efficiency and stability. One difficulty is to maintain a stable morphology of novel donor/acceptor (D/A) blends in highly efficient OPVs, which can easily change even at RT (23, 28, 29). Another difficulty is to fabricate an HTL with stable interfaces among other layers, so that the interfacial morphology and/or properties can be modulated by a high-temperature aging process (18, 19). Any change to the aforementioned layers could lead to dramatic device degradation, and they must therefore be prevented or minimized simultaneously.

Here we develop a simple postannealing method to improve the environmental stability of OPVs without sacrificing their efficiency and demonstrate the combination of this postannealing method with the ultrathin polymer substrate and encapsulation, so as to fabricate ultraflexible OPVs to simultaneously achieve high efficiency, good environmental stability, and areal scalability. Benefiting from the postannealing process, the doping process happens at the interface between the active layer and the molybdenum oxide (MoO_x) HTL, which leads to a more stable interface against high temperatures. With the combination of the thermally stable nonfullerene active layer and the stable interface, the glass-encapsulated annealed devices with a champion PCE of 13.2% achieve good environmental stability with 4.8% efficiency loss after 4,736 h in the dark with air at RT and 20% efficiency loss after 1,050 h at 85 °C and under 30% relative humidity (RH). Furthermore, we extend this method for the large-area fabrication of ultraflexible OPVs with high efficiency and good environmental stability for their application as flexible power sources for wearable electronics.

Results

Device Structure and Environmental Stability.

Fig. 1A shows the OPV structure with rigid glass encapsulation. In this device, a D/A blend of PBDTTT-OFT/IEICO-4F (8, 30) (SI Appendix, Fig. S1) and MoO_x are utilized as the active layer and HTL, respectively. After the device fabrication process, we annealed the devices in an N₂-filled glove box for only 5 min. Interestingly, the OPVs exhibit an unusual change in their performance after annealing at different temperatures (Fig. 1B). Before annealing, the devices have an average PCE of 13.0% (Table 1), short-circuit current density (J_SC) of 25.5 mA/cm², open circuit voltage (V_OC) of 0.698 V, and fill factor (FF) of 73.4%. Although the J_SC decreased to 18.2 mA/cm² after annealing at 90 °C for 5 min in a nitrogen atmosphere, it increased to 26.6 mA/cm² after annealing at 150 °C for 5 min (Table 1). This change in J_SC was confirmed by both external and internal quantum efficiency (EQE and IQE, respectively) measurements (Fig. 1C and SI Appendix, Fig. S2). The quantum efficiencies, EQE and IQE, changed uniformly in the whole range of the responsive wavelengths after annealing. The current densities calculated from EQE measurements are 23.2, 16.3, and 24.1 mA/cm² for the as-fabricated device and devices annealed at 90 and 150 °C for 5 min, respectively, with ∼9% difference in J_SC in J–V curves for all conditions. In sharp contrast to J_SC, both V_OC and FF remained constant after such postannealing treatments. Although the device performance decreased dramatically after annealing at 90 °C, its PCE could be recovered to 13.0% again after annealing at 150 °C.

Fig. 1.

Characterization of nonfullerene OPVs with high efficiency and good environmental stability. (A) Stack structure of the OPV. (B) Comparison of current density–voltage curves and (C) EQE–light wavelength curves of OPVs before and after the single-step postannealing treatment at 90 and 150 °C for 5 min. (D) Evolution of PCE with time when the devices were stored in the dark with air at RT. (E) Comparison of the thermal stability of devices before and after the postannealing treatment at 150 °C before the stability test. The devices were stored in a commercial environment box at 85 °C and 30% RH.

Table 1.

Comparison of the performance of ultraflexible OPVs before and after the postannealing treatment (the effective area is 0.04 cm²)

Tanneal, °C	JSC, mA/cm2	Current, mA	VOC, V	FF	PCE	Power, mW
—	25.5 ± 0.5	1.02 ± 0.02	0.698 ± 0.004	73.4 ± 0.4	13.0 ± 0.2	0.52 ± 0.01
90	18.2 ± 0.4	0.73 ± 0.02	0.695 ± 0.004	72.2 ± 0.4	9.0 ± 0.1	0.36 ± 0.01
150	26.6 ± 0.3	1.06 ± 0.01	0.691 ± 0.003	70.8 ± 0.3	13.0 ± 0.1	0.52 ± 0.01

T_anneal: annealing temperature; the OPVs were annealed at each temperature for 5 min in an N₂-filled glove box.

We then tested the environmental stability of the glass-encapsulated OPVs after postannealing at 150 °C for 5 min. When stored in the dark with air at RT the devices did not degrade within 2,000 h and only lost 4.8% of the initial efficiency after 4,736 h (Fig. 1D) with an estimated T₈₀ lifetime (80% of the initial PCE) of over 20,750 h (SI Appendix, Fig. S3), which will potentially lead to an ultralong shelf lifetime of more than 11.4 y. Furthermore, devices stored in an environmental test chamber at 85 °C and 30% RH showed a T₈₀ lifetime of 1,050 h (Fig. 1E). The degradation mainly arises from the reduced FF (SI Appendix, Fig. S4). Although it is difficult to compare the stability of our device with those reported previously for other devices owing to the variations in the test conditions, the stability of our OPVs is outstanding for a device with an initial PCE of >12% (SI Appendix, Table S1 and Fig. S5). In contrast, under the same test conditions, in the absence of postannealing treatment at 150 °C, our device lost 31.1% of the initial efficiency after 169 h (Fig. 1E). These results clearly demonstrate that the thermal stability of our OPVs is improved significantly after the postannealing treatment at 150 °C, and the related mechanism will be discussed later.

Furthermore, we tested the stability of the devices without any encapsulation and found a 10.9% decrease after 1,394 h of storage in the dark with air at RT (SI Appendix, Fig. S6A). The observed stability is better than that of the state-of-the-art P3HT-based cells with an initial PCE of 7.7%, which show a T₈₀ lifetime of 1,200 h under the same storage condition (25). We also examined the thermal stability of our device in air without encapsulation. PCE retentions of 99.6% and 94.9% were observed for devices heated at 120 °C for 1 and 2 h, respectively (SI Appendix, Fig. S6B). Such good stability in ambient air is advantageous for the processing of OPVs by many other techniques, such as all-solution processing of OPVs (31) and integration with other devices on the same substrate (1). In sharp contrast to the good stability of the PBDTTT-OFT/IEICO-4F OPVs, faster degradation is observed for both PBDTTT-OFT/PC₇₁BM blend and PBDTTT-EFT(PCE10)/IEICO-4F blend (SI Appendix, Figs. S7 and S8). Therefore, we conclude that both stable donor polymers and nonfullerene acceptors are essential for achieving good environmental stability.

We further measured the operation stability, which is very important for OPVs. The measurement details can be found in Materials and Methods. Under the 1-sun condition, our rigid devices only lose 14.2% of the initial value after operating at the maximum output power for 137 h (SI Appendix, Fig. S9).

Mechanism of the Postannealing Effect on OPVs’ Performance.

To investigate the unique effects of the postannealing treatment on the device performance we heated the same device successively at different temperatures between 50 and 200 °C under nitrogen. Since we annealed the same device at each temperature for 5 min, this process is only for investigation of the abnormal performance. As shown in Fig. 2 A–D, the OPVs degraded rapidly upon heating at temperatures between 50 and 90 °C, mainly due to the decrease in J_SC (SI Appendix, Table S2). Surprisingly, the device performance improved dramatically when heated between 90 and 140 °C, and the J_SC increased relative to the initial value. Owing to the slightly decreased FF, the PCE only recovered to ∼90% of the initial value. The same device degraded slowly when heated at temperatures higher than 150 °C. Next, different devices were heated at a given temperature for different durations up to 23 h (SI Appendix, Fig. S10). Both the devices annealed at 85 and 130 °C first showed a decrease in performance, followed by an increase. The performance recovered to 84.1 and 96.8% of the initial value after annealing at 85 °C for 23 h and at 130 °C for 30 min, respectively. These results indicate that the device performance changes significantly with the postannealing condition.

Fig. 2.

Fabrication of nonfullerene OPVs via effective postannealing treatment. (A–D) Evolution of normalized PCE, J_SC, FF, and V_OC with postannealing temperature. Note that the same device was heated from 50 to 200 °C; it was heated at each temperature for 5 min in a nitrogen atmosphere. (E) Comparison of the series resistance and shunt resistance during the postannealing treatment. (F) Comparison of the short circuit current as a function of light intensity during the postannealing treatment. (G) Comparison of transient photocurrent decay curves acquired at 0 V during the postannealing treatment.

To unravel the mechanism of such variation under the three conditions shown in Fig. 1, we considered the differences in the series resistance (R_s) and shunt resistance (R_sh), charge recombination, and charge extraction of the devices (Fig. 2 E–G). We extracted R_s and R_sh from the J–V curves (32). The initial R_s of the device (1.7 Ω∙cm²) increased to 3.5 Ω∙cm² after annealing at 90 °C and then decreased to 2.8 Ω∙cm² after annealing at 150 °C, while the initial R_sh (822 Ω∙cm²) decreased to 300 Ω∙cm² after annealing at 90 °C and then increased to 571 Ω∙cm² after annealing at 150 °C (Fig. 2E). Such variation in resistance is consistent with the variation in the device performance. We investigated the charge-recombination behaviors qualitatively by evaluating the dependence of J_SC with light intensity (Fig. 2F). Under all three conditions, the devices showed a linear power-law dependence of J_SC with light intensity, I $\left(J_{\text{SC}}\propto I^{\alpha }\right)$ (33), with α values of 0.998, 0.987, and 0.996 for the untreated device and devices annealed at 90 and 150 °C, respectively. The α values of the untreated device and that annealed at 150 °C are closer to 1, suggesting negligible bimolecular recombination at the short circuit condition in these devices (33). The differences in charge-extraction dynamics under all three conditions were studied by transient photocurrent (34) measurements (Fig. 2G). The photocurrent decay time increased after annealing at 90 °C and then decreased after annealing at 150 °C as compared to the initial value. The results indicate that the charge extraction in OPVs became less efficient after annealing at 90 °C and much more efficient compared to that of the initial device after annealing at 150 °C.

To elucidate the origin of the thermal annealing-induced change in device performance, we investigated the roles of different materials by systematically changing only one layer or one material in the device. Upon only replacing the ZnO/PEIE bilayer with a ZnO single layer (SI Appendix, Fig. S11 and Table S3), or replacing PBDTTT-OFT/IEICO-4F with PBDTTT-EFT/IEICO-4F (SI Appendix, Fig. S12 and Table S4) or PBDTTT-OFT/PC₇₁BM (SI Appendix, Fig. S13 and Table S5), a similar anomalous change in the efficiency was observed, yet with the highest recovered efficiencies of 11.6, 10.0, and 5.0%, respectively. Finally, we replaced MoO_x with PEDOT:PSS and achieved 11.7% PCE with 66% FF (SI Appendix, Table S6). Interestingly, the decrease and recovery phenomena were not observed during the postannealing treatments (SI Appendix, Fig. S14). These results indicate that the anomalous change in the performance could be related to the changes in the MoO_x layer during thermal annealing. Furthermore, we tested the reproducibility of such abnormal phenomena in OPVs with the MoO_x HTL by changing the thickness and source of MoO_x (SI Appendix, Fig. S15).

To investigate the change related to MoO_x, we first characterized its composition, work function, and bandgap on ITO glass by X-ray photoelectron spectroscopy (XPS), UV photoelectron spectroscopy (UPS), and UV-visible (UV-vis) spectroscopy (SI Appendix, Fig. S16). Differing from previous reports (35, 36), we found no difference in the properties of MoO_x after annealing, indicating that MoO_x itself was stable after annealing.

We then characterized the interfacial property of the ITO/active layer/MoO_x structure. We compared the UV-vis absorption spectra of the active layer and active layer/MoO_x bilayer before and after annealing (Fig. 3A). For the active layer/MoO_x bilayer, the light absorption in the range of 600 to 900 nm decreased after annealing at 150 °C, while that of the pure active layer changed negligibly. The light absorption at ∼726 nm changed negligibly for the active layer bilayer after the annealing process, while that of the active layer/MoO_x bilayer decreased with a blue shift after annealing at 150 °C. Further, there was a very small new absorption peak at ∼1,100 nm after annealing at 90 and 150 °C, which indicates that polaronic or bipolaronic structures formed at the interface between the active layer and MoO_x layers after the postannealing treatment (37). The above phenomenon indicates a change in the interfacial properties during the postannealing process. Fig. 3 B and C show the core-level XPS spectra of S and Mo for the active layer/MoO_x bilayer. Compared with that in the initial state, after annealing at 90 and 150 °C, the S2p peak gradually shifted to a higher binding energy and the Mo⁺⁵ content increased with a concomitant decrease in the Mo⁺⁶ content (36). The UPS results show that the work function of the active layer/MoO_x bilayer shifted from 5.9 eV to 5.6 and 5.5 eV after annealing at 90 and 150 °C for 5 min, respectively (SI Appendix, Fig. S17). We also measured the contact resistance of the devices using the ohmic contact region in the current density voltage curve. For the devices with the film of the pure PBDTTT-OTF, the contact resistance decreased from the initial 16.8 Ω∙cm² to 1.4 and 0.1 Ω∙cm² after annealing at 90 and 150 °C for 5 min (SI Appendix, Fig. S18). A similar phenomenon could be observed for the devices with the film of the PBDTTT-OTF/IEICO-4F.The above phenomenon indicates that the doping process happens between MoO_x and components containing an S atom. Charge transfer occurs between the lone-pair electrons of S and the Lewis acid Mo during the postannealing process (SI Appendix, Fig. S19), leading to the doping process (38). This doping process can also happen to the S atom on other sites of PBDTTT-OFT and even on small-molecule acceptors. The doped interface between the active layer and MoO_x at 150 °C can be more stable than the initial interface, and it contributes to the good device stability. This doping-induced stable interface approach has been utilized for fabricating stable organic light-emitting devices (39) and perovskite solar cells (40–42).

Fig. 3.

Mechanism of the high-temperature tolerance of nonfullerene blends. (A) Comparison of the UV-vis absorption spectra of nonfullerene blends (Top) and nonfullerene blend/MoO_x bilayers (Bottom) before and after the postannealing treatment. It should be noted that the thickness of the active layer is 15 nm for the detection of the new absorption peak at ∼1,100 nm. (B and C) Evolution of XPS spectra of S (B) and Mo (C) at the interface between nonfullerene blends and MoO_x film before and after the postannealing treatment. (D) Comparison of the pc-AFM photocurrent images and topography images of nonfullerene blends captured at 0 V before and after the postannealing treatment. (E and F) Comparison of 1D OOP (E) and IP (F) GIWAXS line profiles of the nonfullerene blends before and after the postannealing treatment.

We speculate that multiple factors cause the abnormal change in the device performance at different temperatures. According to our results, the doping levels between MoO_x and the active layer at 90 and 150 °C are different, which should lead to different levels of enhancement of the device performance (19, 38, 43). The decreased device performance after annealing at 90 °C has to be caused by other factors, such as by a change in the interfacial morphology due to the doping process. Further characterization is needed in the future, which is not within the scope of this work.

Numerous studies have shown that the morphology of the active layer can significantly affect the properties of the device (44). We therefore characterized the surface morphology of the active layers by photoconductive atomic force microscopy (pc-AFM) (45–48). Fig. 3D shows the normalized photocurrent and topography images of the film before and after annealing. The photocurrent of the film annealed at 150 °C increased considerably compared to the initial value, which is consistent with the increased J_SC of the actual device after annealing. The topography images show more pronounced features of the film surface after annealing. We then characterized the crystallinity and packing of the film by grazing incident wide angle X-ray scattering (GIWAXS). Two-dimensional (2D) GIWAXS patterns are shown in SI Appendix, Fig. S20 and their corresponding line profiles along out-of-plane (OOP) and in-plane (IP) are presented in Fig. 3 E and F. The scattering peak located at 1.8 Å⁻¹ along OOP direction correspond to π–π stackings of IEICO-4F, and the scattering peaks located at q = 0.27 and 0.31 Å⁻¹ along the IP direction correspond to lamellar stackings of PBDTTT-OFT and IEICO-4F, respectively (49). There is no obvious change in either the 2D GIWAXS patterns or the extracted peak positions after the postannealing treatment. Since both the surface morphology and the crystallinity and packing of the film (as revealed by Pc-AFM and GIWAXS) show no obvious change, the nonfullerene blends are very stable after heating at 150 °C, which contributes to the good environmental stability of the devices. Pc-AFM images of the blended film annealed at 130 °C for 24 h (SI Appendix, Fig. S21) only show slightly larger features on the surface with no change in the photocurrent as compared to those of the initial film (Fig. 3D). Note that the photocurrent of the fullerene blends decreased dramatically after the same thermal treatment (SI Appendix, Fig. S22).

Ultraflexible Devices with Scalability and Stability.

We then fabricated ultraflexible OPVs with less than 3-µm thickness (Fig. 4). After the standard fabrication, we peeled off the ultraflexible OPVs from the supporting glass substrates (Fig. 4A). The device performance did not degrade after the peeling process (Fig. 4B). Fig. 4B, Inset shows a photograph of the free-standing device rolled around a 75-µm-diameter copper wire. When ultraflexible devices were stored in the dark with air at RT, they only lost 4.3% of the initial efficiency after 3,194 h (Fig. 4C). Such storage stability is also the best for flexible/ultraflexible devices. Compared with the glass-encapsulated devices, the slightly lower stability of the ultraflexible OPVs is due to their insufficient encapsulation in ultrathin polymer films, which provide less protection from oxygen and water than glass (8). A better encapsulation film (50, 51) with good mechanical property can further improve the stability.

Fig. 4.

Ultraflexible and large-area nonfullerene OPVs. (A) Illustration of the peeling process of the device from the supporting glass substrates. (B) Comparison of the current density–voltage curves of OPVs before and after they were peeled off from the supporting glass substrates. (Inset) Optical image of an ultraflexible device. (C) Evolution of PCE with time when the devices were stored in dark in ambient air at RT. (D) Current density–voltage and current–voltage curves of a single subcell of OPV modules. (Inset) Optical image of the mini module with three 0.68-cm² subcells. (E) Illustration of the mechanical test for the ultraflexible devices. A buckling structure can be formed when releasing the prestretched elastomer. (F) Mechanical stability of the OPV module under cyclic compression/extension test with 12% compression. (Inset) Optical image of the minimodule under 12% compression.

Areal scalability is important for achieving a large output power. We designed a minimodule consisting of three subcells with 0.68-cm² area in serial connection (Fig. 4D). In this module, the average PCE of each subcell is 11.7%, with J_SC of 26.2 mA/cm², V_OC of 0.7 V, FF of 64%, and maximum power output of 7.96 mW. The maximum output power of the modules is almost scaled by the number of subcells connected serially (SI Appendix, Fig. S23 and Table S7). After the peeling-off process, ultraflexible minimodules showed 11.4% PCE with a total weight of 2.35 mg and a maximum power per unit weight of 9.90 W/g under a 75.5% aperture ratio. Such a high power per weight can be beneficial to portable devices (SI Appendix, Fig. S24).

We further characterized the mechanical cyclic durability of the ultraflexible OPVs. We transferred the ultraflexible OPVs onto a prestretched adhesive elastomer and then released the elastomer to compress the OPVs to form a buckling structure (Fig. 4E). The small-area ultraflexible OPVs showed 90% PCE retention after 1,000 compression/extension cycles with 37% compression (SI Appendix, Fig. S25). The minor degradation mainly originates from reduced J_SC and FF, possibly caused by the increased defect density during mechanical deformation (10). The large-area module also showed good mechanical stability at a minimum bending radius of less than 100 µm (Fig. 4F and SI Appendix, Fig. S26). The efficiency decreased slightly from 11.4 to 11.0% after the first cycle, and the same value was retained during the subsequent cycles with a compression of 12%. Additionally, the devices could operate stably for 3 h with only 4.8% PCE loss (SI Appendix, Fig. S27).

Conclusions

In conclusion, a simple single-step annealing approach was developed to fabricate OPVs with high efficiency, good stability, and areal scalability. The remarkable performance of the OPVs is due to the synergetic effect of stable D/A blends and doping-induced stabilization of interfaces. The high efficiency and stability are retained in ultraflexible devices and modules, which also have good mechanical stability and long-term operation capability. Our findings demonstrate the promise of flexible OPVs that can be stably used as power sources in wearable electronics.

Materials and Methods

Materials.

A PBDTTT-OFT:IEICO-4F solution (25 mg/mL) was prepared by blending PBDTTT-OFT (Toray Industries) and IEICO-4F (1-Material) in a solution of chlorobenzene (C₆H₅Cl, Wako) with 3% 1-chloronaphthalene additive (Sigma-Aldrich). Similarly, a PBDTTT-OFT:PC₇₁BM solution (20 mg/mL) was prepared by blending PBDTTT-OFT (Toray Industries) and PC₇₁BM (99.5% purity ; Solenne BV Corporation) at a weight ratio of 0.9:1.1 in a solution of chlorobenzene (C₆H₅Cl, Wako) with 2% 1,8-diiodooctane additive (Sigma-Aldrich). Further, a ZnO precursor solution was prepared by dissolving zinc acetate dehydrate (0.55 g; Wako Chemicals) and ethanolamine (0.16 mL; Wako Chemicals) in 5 mL of 2-methoxyethanol (Wako Chemicals). Ethoxylated polyethyleneimine (PEIE) (Sigma-Aldrich) was diluted to 0.4% with 2-methoxyethanol (Wako Chemicals).

Device Fabrication and Characterization.

The OPVs were fabricated by a process similar to that reported previously (9). Bare glass substrates were treated with O₂ plasma for 10 min at 300-W power. Then, they were coated with a ∼1-µm-thick layer of transparent polyimide (ECRIOS; Mitsui Chemicals). The polyimide layer was cured into a film by annealing at 270 °C for 2 h in a N₂ atmosphere. Then, a 100-nm-thick ITO transparent electrode was deposited on each of the substrates through sputtering, and the electrode layer was patterned by photolithography and a wet-etching process. Subsequently, Cr/Au electrodes (3.5-nm-thick Cr and 80-nm-thick Au) were deposited as contact pads. A ZnO–PEIE bilayer formed by spin-coating ZnO precursor (baked at 180 °C for 30 min) and PEIE solution was used as the electron-transport layer. For the formation of active layers, nonfullerene and fullerene blends were spin-coated at 1,400 rpm for 1 min and 600 rpm for 20 s, respectively. Then, MoO_x (7.5 nm; Wako chemcials) and Ag (100 nm) films were deposited by thermal evaporation. Before the devices were encapsulated in a parylene (diX-SR; Daisan Kasei Co., Ltd.) layer by chemical vapor deposition, external wiring (Cr/Au 3 nm/100 nm on polyimide substrates) was connected to the OPVs using electrically conductive adhesive tape (ECATT 9703; 3M Company). For the rigid devices, we cast a thin layer of the epoxy glue on the top of the whole parylene layer and then covered the device with a glass slice. To solidify the epoxy glue, we heated the device at 100 °C for 5 min. For ultraflexible devices, there is no such epoxy glue or top glass encapsulation. Finally, the OPVs were annealed at 150 °C in an N₂-filled glove box for 5 min. OPVs were characterized under 1-sun illumination using a solar simulator (AM 1.5 global spectrum with 1,000 W/m² intensity calibrated with a silicon reference diode). The current density–voltage characteristics were recorded using a Keithley 2400 source meter (Tektronix, Inc.) between −1 and +1 V at scanning increments of 20 mV. For EQE measurement, OPVs with 0.2 cm² effective area were covered by a metal mask of 0.12 cm² size, and the EQE data were acquired with monochromatic light (SM-250F; Bunkoh-Keiki). Transient photocurrent decay curves were recorded using a digital oscilloscope (Infiniivision DSO-X 3032A; Agilent Technologies); the device was connected to a 50-Ω termination resistor to achieve a short-circuit condition.

Mechanical Test.

For mechanical testing of OPVs, we attached the ultrathin OPVs delaminated from supporting glasses onto an elastomer prestretched to 200%. During the release of the prestretched elastomer, the OPVs formed a buckled structure, and their current density–voltage curves under different strain were recorded. Compression was calculated by the areal change determined through laser microscope imaging.

Thermal Stability Test.

To examine the thermal stability following the ISOS-D-2 protocol, the devices were stored in a commercial environment chamber (SH-242 bench-top-type temperature and humidity chamber; Espec) at a controlled temperature of 85 °C and RH of 30%. The shelf life was determined following the ISOS-D-1 protocol, that is, devices were stored in ambient conditions (in a light-shielded desiccator at a controlled temperature of 25 °C and humidity of 30%) and tested periodically under AM 1.5 illumination in ambient atmosphere.

Operational Stability Test.

We used the MPP tracking test method following the International Summit on OPV Stability (ISOS) protocol (ISOS-L-1). We continuously applied voltages at the maximum power point to OPVs under a 1-sun condition. The current density–voltage curves were recorded every 5 min in both forward and backward directions, and the voltage at the maximum power point was updated accordingly. The measurements were performed in the ambient air condition, and the temperature of the bottom black table could increase to a saturated temperature of 42 °C, owing to the 1-sun irradiation without a cooling system.

Characterization.

pc-AFM was performed on an Asylum Research MFP-3D microscope placed atop an inverted optical microscope (IX71; Olympus). All measurements were conducted under inert atmosphere. A white light source with a power of 30 W⋅cm⁻² was used to generate the photocurrent, which was recorded using an internal preamplifier (Asylum Research ORCA head model). Electrically conductive Cr/Pt-coated silicon probes with a spring constant of 0.2 N⋅m⁻¹ and resonant frequency of 13 kHz (Budget Sensors) were used. A light spot with 160 μm diameter was focused on the sample through an inverted optical microscope (Olympus), and the conductive tip was positioned at the center of the light spot. GIWAXS measurements were conducted at the Australian Synchrotron on the SAXS/WAXS beamline. All samples were following the same procedure as device fabrication. Samples were irradiated by 10-keV X-ray beam with different X-ray incident angles varying from 0.1° (below critical angle) to 0.4° (above critical angle). A Dectris Pilatus 1M detector was used to collect the 2D scattering patterns. The X-ray exposure time was 1 s and no film damage was observed. Absorption spectra in the transmission mode were acquired with a UV-vis spectrophotometer (V-670; Jasco). XPS was performed with a surface analysis instrument (PHI 5000 VersaProbe II; ULVAC-PHI). Monochromatized Al-Kα (photon energy, 1486.6 eV) radiation was used for all of the measurements. To measure the interface between the MoO_x layer and active layer, a 1.2-nm MoO_x layer was deposited on the active layer on ITO-coated glass substrates. UV photoelectron (UPS) spectroscopy was performed on a surface analysis instrument (PHI5000 VersaProbe II; ULVAC PHI Inc.) with He(_I) excitation (photon energy, 21.2 eV). For all UPS measurements, samples were prepared on ITO substrates and a 5-V bias was applied.

Data Availability Statement.

All data discussed in the paper are available in SI Appendix and Dataset S1.