A generic approach yields organic solar cells with enhanced efficiency and thermal stability

The use of deuterium are critical for promoting the fundamental understanding of aggregate materials and their new functions. Particularly, the solution structure of conjugated polymers can be hardly resolved without deuteration. However, studies about the isotopic effects of casting solvents on the aggregated structures of photovoltaic polymers and their bulk‐heterojunction blends are deficient. Here, the impact of deuterated solvents on the thermal behavior, aggregated structures, and device performance of photovoltaic polymers is clearly delineated for the first time by multiple techniques. The enhanced π‐π stacking order of photovoltaic polymers is highly relevant to their relatively poor miscibility with deuterated solvents. Benefiting from higher crystallinity and optimized morphology of deuterated solvents processed films, the devices are able to achieve better efficiency and notable improvement in thermal stability. Our results highlight the isotopic effects of solvents on the aggregated structure of conjugated polymer systems and reveal the potential of innovative approaches to fabricate thermally stable high‐efficiency solar cells.

The scattering lengths of two hydrogen isotopes, protium and deuterium, where the protium atom has a negative scattering length due to incoherent scattering. The scattering length difference between protium and deuterium tremendously empowers the study of polymer structure by SANS. Clearly, the scattering contrast of SANS can be enhanced by replacing the undeuterated solvent with its deuterated version. (C) Plots of the scattering contrast (Δρ) between representative photovoltaic polymers and chloroform (CF), deuterated chloroform (CF-d), o-dichlorobenzene (DCB), and deuterated o-dichlorobenzene (DCB-d), respectively. The Δρ value should be above 2 × 10 −6 Å −2 (above the gray plane in the figure) to obtain valuable data with a sufficient signal-to-noise ratio. (D) The SANS profiles of a representative photovoltaic polymer in DCB solvent and DCB-d solvent, respectively. (E) Chemical structures of representative polymers and the schematic of the device structure of OSCs are fabricated by CF-d or DCB-d solution.
(such as π-π interaction, and hydrogen bonding) between materials and solvent induce it difficult to characterize and control the solution-state aggregated structure. [23,24] In the past decade, the control and understanding of polymer microstructure rely heavily on advanced characterization tools such as conductive atomic force microscopy, cryogenic transmission electron microscopy, [25] grazing incidence wide-angle X-ray scattering (GIWAXS), [26] resonant soft X-ray scattering, [27] small-angle neutron scattering (SANS), [28] etc. To date, SANS is one of the very few techniques (schematically shown in Figure 1A) that can resolve the solution structure of polymer chains, facilitating the establishment of the relations between solution-state and solid-state structure, and then accurately guiding the morphological optimizations and device perfor-mance improvements. [28][29][30] Due to the significant difference between the scattering lengths of two hydrogen isotopes, protium and deuterium, the scattering contrast (i.e., the difference in scattering length density, Δρ) for SANS measurement can be enhanced by replacing protium with deuterium (see Figure 1B). Also, the incoherent scattering background from protium can be lowered by using deuterium, which is critical to obtaining high-quality SANS data. We first evaluated the Δρ of a wide range of representative conjugated materials (16 photovoltaic polymers and 22 nonfullerene small molecule acceptors, see Figure S1) with the two most commonly used and representative solvents for device fabrication, namely o-dichlorobenzene (DCB) and chloroform (CF), as shown in Figure 1C and Figure S2. Unfortunately, the Δρ values of these representative photovoltaic materials and common casting solvents are generally lower than 2 × 10 −6 Å −2 , which makes it challenging to analyze the SANS data. As displayed in Figure 1D, the low contrast of a representative conjugated polymer in DCB severely precludes the use of SANS to characterize its solution structure of OSCs. Therefore, deuteration is a quite effective means to accurately reveal the solution structure of conjugated polymers.
Herein, we put forward a conceptually simpler and generic approach to address this challenge without the tedious deuteration of conjugated polymers. By using deuterated organic solvents, deuterated chloroform (CF-d) and deuterated o-DCB (hereafter named DCB-d) as casting solvents, we have been able to decipher the isotope effect of casting solvents on photovoltaic polymer systems from the aspects of thermal properties, molecular stacking, morphology, and device properties. We proved that deuterated solvents can facilitate the characterization of the solution structure of photovoltaic polymers and obtain data with a much higher signal-to-noise ratio (see Figure 1D). The effects of deuterated solvents (CFd and DCB-d, see Figure 1E) on morphology and device performance were revealed with fast-scanning calorimetry, X-ray scattering, neutron scattering, and microscopic characterizations by taking the high-performance polythiophene derivative PTVT-T as the model system. Our results show that the π-π interactions of conjugated polymers in deuterated solvents are enhanced, leading to higher crystallinity and more ordered molecular stacking, which is conducive to exciton dissociation and charge transport. Importantly, we discovered that deuterated solvents cannot only maintain the device performance but also promote the thermal stability of OSC active layers. For example, the T 80 lifetime of PTVT-T:BTP-eC9 solar cells processed by CF-d is more than eight times of those processed by CF. Similar improvements in PCE are further observed in benchmark P3HT:O-IDTBR [31] and PM6:N3 systems. [32,33] The blend films processed by DCB-d also displayed higher crystallinity and thus have more efficient carrier transport. This work reveals another potential for deuterated solvents, namely optimizing film morphology and fabricating thermally stable high efficiency solar cells.

Aggregation and crystalline structure
First, the effect of solvents with isotopic substitution on the optical properties was characterized by UV-Vis absorption spectroscopy (UV-Vis). Figure S3A depicted that PTVT-T had completely consistent solution absorption spectra with J-type aggregates in CF and CF-d solution, exhibiting an intense 0-0 transition peak at a higher wavelength (λ 0-0 = 625 nm) with a shoulder-type absorption peak at a lower wavelength (λ 0-1 = 576 nm). Normalized UV-Vis absorption spectra of neat PTVT-T thin films fabricated out of CF and CF-d are shown in Figure S3B. The λ 0-0 and λ 0-1 absorption peaks of PTVT-T thin films were red-shifted to 648 and 589 nm, respectively. In the films processed by two solvents, the absorption peaks have the same peak position, while the intensities of absorption peaks are slightly different. The ratios of (0-0) and (0-1) peak intensity (I( 0-0) /I (0-1) ) are 1.22 and 1.25 in the thin films processed by CF and CF-d, respectively. These differences demonstrate that the CF-d processed film exhibited slightly stronger π-π intermolecular interactions.
To investigate the effect of deuterated solvent on the crystallization behavior of photovoltaic polymer, quantification of thermal properties and nanoscale ordering features by advanced calorimetry and X-ray scattering is critical. Herein, fast scanning calorimetry (FSC) was utilized to accurately distinguish the crystallization behaviors of polymer thin films. FSC has the merits of higher sensitivity, fast scanning rates (up to 40000 • C/s), and a wide scanning temperature range (from −95 • C to 1000 • C). These features allow the use of nanoscale samples and such a fast heating rate prevents the thermal degradation of polymers. Figure  . The generality of this phenomenon was verified by conventional DSC in conjugated polymers with stronger crystallinity, such as poly (3-octylthiophene), [34] as shown in Figure S4. In short, the above results indicated that the CF-d-derived films have higher crystal perfection and more stable crystalline forms.
Further, GIWAXS measurements were carried out to clarify the effects of deuterated solvents on molecular stacking and crystallinity. Figure 2B displayed the 2D diffraction patterns of PTVT-T neat films processed by CF and CFd, respectively. The film processed by CF-d showed a stronger (010) diffraction signal in the out-of-plane (OOP) direction, while the film processed by CF depicted a diffraction ring, indicating that the CF-based film appears to be bimodally oriented with poor crystallinity (schematic diagram in Figure 2B). The 1D scattering profiles are depicted in Figure 2C. CF-d processed films had stronger π-π stacking of backbone and a higher degree of self-organization, which can be indicated by the larger values of coherence length (CL), and the π-π stacking peak area of CF-d film is about 1.7 times that of CF film, as reflected in Figure 2D. The above phenomena suggest that the use of CF-d processing facilitates the formation of more ordered π-π stacking in the PTVT-T film. To quantitatively analyze the crystallization properties of PTVT-T in deuterated solvents, the pole figures of (100) diffraction peaks of PTVT-T neat films processed by CF and CF-d ( Figure 2E) were employed to acquire the relative degree of crystallinity (rDoC). The detailed analysis procedures are presented in our previous work. [35] As depicted in Figure 2F, the rDoC of PTVT-T in CF-d was higher than that of PTVT-T in CF, indicating the lamellar packing also took a crucial role in forming the crystalline structure of PTVT-T processed by CF-d. In addition, peak force quantitative nanomechanical mapping was employed to characterize the elastic modulus of these neat films processed by different solvents. [36] As shown in Figure 3A,B, the average modulus of PTVT-T film CF (0.47 ± 0.04 GPa) was slightly lower than that in CF-d (0.51 ± 0.03 GPa). The mechanical properties of PTVT-T films in accordance with the GIWAXS result pointed out that the CF-d processed film has higher crystallinity due to the strong π-π interaction and ordered side chains. Therefore, we experimentally demonstrated that the deuteration of solvent affects the molecular stacking and crystallinity of the photovoltaic polymer, which may further lead to differences in film morphology.

Morphological features
We further investigated the effect of deuterated solvent on the mesoscale morphology of PTVT-T films by atomic force microscopy (AFM) and transmission electron microscopy (TEM). PTVT-T neat films presented the clearly fibrillar aggregates with the mean square roughness (R q ) of 1.3 nm and 1.4 nm for the films processed by CF and CF-d, respectively. We statistically averaged the R q of films to ensure the reliability of the differences. To quantitatively evaluate the fibril width of these films, we calculated the statistical average of the fiber widths based on AFM height images ( Figure 3 and Figure S5). The statistical results show that the average fiber width of the film processed with CF are about 50 nm, which is 1.4 times that of the fiber width processed with CF-d (36 nm). TEM images in Figure 3 clearly show that the films spin-coated by CF and CF-d both formed ordered fibrillar aggregates. Then, the power spectral density (PSD) analysis of the TEM images was employed to extract the characteristic length scale ( Figure 3C). Combined with the TEM images, the main characteristic peak may correspond to the fiber width. The fiber width characteristic size (s = 2π/q) of the film processed by CF-d (33 nm) was smaller than that by CF (48 nm), which are consistent with AFM height images analysis.

Photovoltaic performance and thermal stability
To understand the impact of deuterated solvents on photovoltaic properties, we fabricated the devices with a configuration of indium tin oxide (ITO)/PEDOT:PSS/PTVT-T:BTP-eC9/PFN-Br/Ag, and the detailed preparation processes for OSCs are shown in supporting information. The devices were fabricated from CF and CF-d solutions using identical conditions for comparison purposes. Figure 4A depicts the current density versus voltage (J-V) curves of devices under AM 1.5 G, 100 mW cm −2 irradiation, and the device parameters are summarized in Table 1. According to the J-V curves, the best performing device based on CF as the solvent delivered an optimized PCE of 14.03% with a V OC of 0.81 V, a J SC of 23.85 mA cm −2 , and an FF of 72.6%. While the optimized photovoltaic devices processed with CF-d attained an enhanced PCE of 14.96% with an increased FF of 74.7%, a J SC of 24.73 mA cm −2 , and a V OC of 0.81 V. As displayed in Figure 4B, the external quantum efficiencies are in agreement with the J SC from J-V tests. We also note that the device fabricated by CF-d delivers a more efficient photon response in the range from 550 to 700 nm, which may be due to the efficient exciton dissociation and charge transport. Since the main absorption peak of PTVT-T in the range of 550-700 nm, we suspect that deuterated solvent processed films have higher charge transfer efficiency. Therefore, the space-charge limited current was used to evaluate the hole mobilities (μ h ) of PTVT-T neat films in the two solvents. The μ h of PTVT-T neat film prepared with CF and CF-d were 7.3 × 10 −4 cm 2 V −1 s −1 , and 1.1 × 10 −3 cm 2 V −1 s −1 , respectively ( Figure S6). Similarly, the μ h and electron mobility (μ e ) of the blend films prepared by CF-d have higher mobilities. Besides, well-balanced (μ h /μ e = 1.05) charge mobilities ( Figure 4C) are observed, while the μ h /μ e ratio of blend films processed by CF is 1.20. The high and balanced charge mobilites facilitate the charge extract and transport in the active layer, hance the slightly higher FF and J SC of the corresponding devices.
Hoping to further elucidate the role of deuterated solvents on the operation lifetime of OSCs under thermal stress, we focused on the thermal stability of devices. Figure 4D depicts the normalized PCE and extrapolated T 80 versus the annealed time of the active layers fabricated by CF and CF-d under continuous thermal annealing at 85 • C in the nitrogen atmosphere under the dark condition. It is obvious that the CF-d-based devices could maintain over 95% of the original PCE after the active layer annealed for 1000 h and exhibited an extrapolated T 80 lifetime of up to 8000 h, which is among the best values for OPV active layers under heating at 85 • C. By comparison, the T 80 of CF-based devices under the same annealing conditions is about one-ninth of that of CF-d-based devices. The main reasons for the poor thermal stability are the gradual decrease in V OC and FF with the longer annealing time ( Figure 4E). Excitingly, we discover for the first time that the utilization of deuterated solvents instead of conventional solvents can not only achieve an enhancement in the photovoltaic performance of OSCs but also significantly extend their thermal stability.

Optimal morphology and its evolution
Generally, the photovoltaic performance and thermal stability of OSCs are correlated to the molecular stacking structures and morphological features of the active layers. [35,37,38] To elucidate the difference in performance and thermal stability between the two processing solvents, the crystalline structure and morphology of blend films were characterized by GIWAXS, AFM, and TEM. As shown in Figure 5A,B, the PTVT-T:BTP-eC9 blend films processed by CF and CF-d both exhibited predominantly face-on orientation, evidenced by the quite strong (010) π-π stacking peaks in OOP direction and (100) lamellar stacking peaks in IP direction in 2D diffraction patterns and 1D line-cut profiles. The π-π stacking peaks of PTVT-T and BTP-eC9 were merged together and indistinguishable with the peak location at ∼1.75 Å −1 , which is evidence of strong interactions between the donor and acceptor. Detailed multipeak fitting curves and fitting results are depicted in Figure S7 and Table S1. As displayed in Figure S8 and Figure 5C, the face-on/edge-on fractions of (100) diffraction peaks are also investigated. The film processed by CF-d exhibited a higher percentage of face-on orientation (73%) than CF-derived film (62%). Moreover, the blend film processed by CF-d had a higher rDoC value, indicating that more favorable charge transport was realized in the CF-d processed blend film; this results in a higher J SC and FF of the devices. After blending the acceptor BTP-eC9 with PTVT-T, the morphology changed from a fibrillar aggregated structure to obvious particle aggregation due to the high miscibility of PTVT-T and BTP-eC9. The R q of blend film processed by CF-d (5.6 nm) was higher than that by CF (4.7 nm), which can be attributed to the enhanced crystallinity of CF-d-processed blend film ( Figure 5D). The TEM images under optimized conditions are shown in Figure 5E (left). It can be seen that the blend film processed by CF-d had clear phase interfaces, slightly higher contrast, and integrated intensity from TEM PSD (see Figure 5F and Table S2), which may imply the higher purity of CF-d processed films. [39] It has been demonstrated that under the condition of ensuring the same incident photon energy, mass/electron density correlations, and electron diffraction contrast, the integrated intensity from TEM PSD profiles could characterize the phase purity of the films to a certain extent. [39] Based on the higher rDoC, the larger surface roughness, and higher contrast of TEM images, we speculate that the films processed with deuterated solvents have higher crystallinity and phase purity, which likely promote efficient exciton dissociation and charge transport. For the long-term annealed (up to 1000 h) films ( Figure 5E, right), the normalized integrated intensities of these films were estimated in Table S2. After long-term thermal annealing, the characteristic peak integral intensity of the CF-d processed film has been significantly improved while the peak position has little changed, indicating that the phase purity of the blend film has been greatly enhanced with a stable domain size ( Figure 5F). However, the characteristic peak of CF processed films moved to high q after long-term thermal annealing, with the characteristic size of the phase domains decreasing from 161 to 143 nm with the increased phase interface.
To further reveal the origins of the higher thermal stability of PTVT-T:BTP-eC9 processed with CF-d, the UV-Vis absorption spectra of the blends were recorded as a function of annealing time ( Figure S9A,B). The absorption peaks at 648 and 827 nm were assigned to donor and acceptor absorption peaks, respectively. Figure S9C displayed the intensity ratio of absorption peaks of the donor and acceptor (I D /I A ) versus the annealing time of the active layers under continuous thermal annealing at 85 • C. The intensity of the donor absorption peak decreased slightly at first and then tended to be stable with the extension of annealing time in CF-d films.
In contrast, the film processed with CF showed an unstable variation of absorption peak intensity, indicating the instability of molecular stacking during thermal annealing. We thus ascribe the enhanced thermal stability in the devices processed with deuterated solvent to the kinetic isotope effect and optimized bulk morphology, which will be discussed below.

Isotope effects of deuterated solvents
The above results conclusively show that the isotope effect of solvents has a clear influence on the crystallinity of conjugated polymer thin films and thus induces the difference in photovoltaic properties of OSCs. Regarding the essence of isotopic substitution, deuterium substitution usually reduces the molar volume and polarizability of the molecules by decreasing the bond length and dipole moment of C─H bond. [40] The relative magnitudes of these changes would decrease the solubility parameter (δ) with deuterium substitution. In 1975, Henri Benoit et al. [41] determined the thermodynamic properties of poly(protostyrene) in cyclohexane (C 6 H 12 ) and deuterated cyclohexane (C 6 D 12 ) by means of light scattering from the temperature dependence of the second virial coefficient A 2 . The experimental results show that the PS-C 6 D 12 system had higher critical miscibility temperatures θ, indicating the lower δ for the deuterated solvent. For a polymer-solvent system, the miscibility of this system can be quantified by the Flory-Huggins parameter χ, that is, = s + V 1

RT
( 2 − 1 ) 2 , the δ 1 and δ 2 are the solubility parameters of solvent and polymer, respectively; χ s is the entropic contribution to the parameter χ, which is usually taken as an empirical value of 0.34. [42] Deuterium substitution decreases the δ 1 , which in combination with the above equation yields an increase in the χ. Thus, the deuterated solvent-polymer system is less miscible compared to the undeuterated solvent-polymer system. A broad program by Van Hook's group [43] also confirmed that the χ parameter of deuterated solvent-polymer is higher.
Here, we note that PTVT-T has a rigid molecular chain with a large Kuhn length of ∼24 nm, as determined by SANS ( Figure S10), as the dihedral angle of thiophene and vinyl in TVT unit on the backbone is about zero and the O⋅⋅⋅H noncovalent interaction is strong. [29,44] Therefore, the low miscibility of polymer and deuterated solvent easily induces enhanced π-π interactions of the conjugated polymer backbone, which results in a higher nucleus density in solution, reduces the energy barrier of nucleation, eventually leading to an ordered molecular packing and a highly crystalline film with purer phase. The optimized film morphology promotes the dissociation of excitons and the transport of charge carriers, inducing the higher PCEs of deuterated solvent processed devices. It is worth mentioning that the molecular stacking structure and morphology are maintained under the long-term thermal annealing, which is also probably due to the strong π-π intermolecular interactions that allow the film to form a tighter molecular stacking at the initial state. At the same time, the amorphous molecular chains move locally under thermal stress, thereby improving the phase purity while keeping the domain sizes basically unchanged, as well as inducing the devices processed by deuterated solvent highly stable against heating (as shown in Figure 6A). In contrast, the weaker aggregation of conjugated polymers in undeuterated solvents and the high miscibility between donor and acceptor together result in films with poorer crystallinity and phase purity. Therefore, the molecular rearrangement and crystallization lead to increased crystallinity and more phase interfaces in the films after long-term thermal annealing. Besides, the phase purity of the amorphous mixed domain remains poor (see Figure 6B), which inhibits the  Figure 4E).

Broad applications of deuterated solvents
To illustrate the potential applicability of deuterated solvent for optimizing morphology and improving the device performance and thermal stability of OSCs, we further explored the isotope effect of solvent with other blend systems and solvents, all showing that deuterated solvent can enhance the device performance ( Figure 7A). First, the benchmark combination PM6:N3 of remarkable interest was selected as an additional system. As shown in Figure 7B,C and Figure S11, higher rDoC values are observed in the CF-d derived film by GIWAXS. Furthermore, the blend films processed with deuterated and undeuterated solvents have similar characteristic length scales, while the CF-d processed films have a higher integrated intensity ( Figure 7D), indicative of a higher phase purity. As a consequence, enhanced PCE was recorded in CF-d processed OSCs ( Figure 7E, Table 1). We have also verified the effectiveness of deuterated solvents to enhance device performance in the classical P3HT:O-IDTBR system, which also yields a promising efficiency of over 7.69% (see Figure 7F, and Table 1). We note that 7.69% is above the prior record for P3HT:O-IDTBR system. [45] Besides, the positive effect of this approach has also been demonstrated in state-of-the-art all-polymer solar cells, such as PM6:PYF-T-o system [46] (Figure S12, Table 1). Lastly, we extended our research to a different pair of solvents, DCB and DCB-d. The results demonstrated the ability of deuterated solvents to improve the crystallinity and surface roughness of blend films ( Figure 7G,H), and thus the PTVT-T:BTP-eC9 devices employing by DCB-d solvent afforded a higher J SC and PCE ( Figure 7I) Figure S13). By extrapolating the T 80 lifetime from the available thermal stability data, we found that the deuterated solvents processed films all exhibit supe-rior thermal stability. Therefore, the use of deuterated solvent can not only achieve the enhancement in device performance of many kinds of OSCs but also improve the thermal stability of devices.
Significantly, our approach of applying deuterated solvents may be capable of offering advantages to several fields and deserves in-depth understanding. For example, Solanki et al. found that the PCE of perovskite solar cells can be enhanced with reduced trap-assisted recombination by utilizing heavy water as a solvent additive in a pre-dissolved perovskite solution. [47] Besides, the isotope effect has also been proven to have a positive effect on the performance of luminescent covalent organic frameworks [48] and organic light-emitting small molecules. [49] Importantly, it is also more conducive to establishing the solution structure-performance frameworks based on the study of the solution aggregated structure, which will considerably benefit from the improvement of scattering contrast in SANS measurements by utilizing the deuterated solvents. Besides, revealing the solvent isotope effect on deuterated photovoltaic materials [50,51] is a direction worth exploring in depth.

CONCLUSION
In summary, we demonstrate for the first time the unexplored potential of deuterated solvents for optimizing molecular stacking and phase morphology with appreciable device performance and superior thermal stability in OSCs. The use of deuterated solvent instead of conventional solvent as the processing solvent of OSCs based on PTVT-T:BTP-eC9 achieved a slight enhancement in PCE and an obvious improvement in thermal stability with an approximately 8fold increase in T 80 . From our experimental findings, the deuterated solvent is beneficial in several ways: (i) the poor miscibility between deuterated solvent and polymer enhances the π-π interactions of conjugated polymer; (ii) improved film quality mainly attributed to the ordered molecular stacking and enhanced rDoC; and (iii) the active layer presented purer phase and stable morphology are conducive to efficient charge transport and the rise in FF, as well as the thermal stability. Our research not only reveals the feasibility of using deuterated solvents to process OSCs but also unravels its role in improving the device performance of OSCs, which has been verified in three kinds of photoactive material systems (PM6:N3, P3HT:O-IDTBR, and PM6:PYF-T-o) and using various deuterated solvents. This is a previously unexplored and facile approach to controlling and improving the photovoltaic properties of OSCs. For all-polymer solar cells where morphology is difficult to be tuned, [52,54] the selection of a suitable deuterated solvent as the processing solvent may be able to address the difficulties of morphological characterizations in solution. Desipte the potential cost and scarcity concerns of deuterated solvents, their utility is still worthy of further explorations in many other kinds of polymer electronics such as photodetectors, transistors, and light-emitting diodes, especially from a scientific standpoint.  1515110984). The authors are also grateful for access to the beamline BL16B1 and BL02U2 of Shanghai Synchrotron Radiation Facility (SSRF) and the beamline 1W1A of Beijing Synchrotron Radiation Facility (BSRF) for supporting the two-dimensional GIWAXS measurements. The SANS experiments were performed on the small-angle neutron scattering instrument at the China Spallation Neutron Source (CSNS), Dongguan, China (project ID: P0121122200038). The SANS instrument maintenance/ support by Dr. Yubin Ke and staffs are greatly appreciated.

C O N F L I C T O F I N T E R E S T
The authors declare no conflict of interest.

D ATA AVA I L A B I L I T Y S TAT E M E N T
The data that support the findings of this study are available from the corresponding author upon reasonable request.