11.4% Efficiency non-fullerene polymer solar cells with trialkylsilyl substituted 2D-conjugated polymer as donor — TEI-tagged view

Haijun BinpersName, Liang GaopersName, Zhi-Guo ZhangpersName, Yankang YangpersName, Yindong ZhangpersName, Chunfeng ZhangpersName, Shanshan ChenpersName, Lingwei XuepersName, Changduk YangpersName, Min XiaopersName, Yongfang LipersName (2016date). “11.4% Efficiency non-fullerene polymer solar cells with trialkylsilyl substituted 2D-conjugated polymer as donor.” Nature Communicationstitle·j.

Object of study — a trialkylsilyl substituted 2D-conjugated donor polymer (J71) in non-fullereneterm·key polymer solar cellsterm·dom

Research question — how to obtain high open circuit voltageterm·key and high short circuit current at once in polymer solar cellsterm·dom

Contribution — alkylsilylterm·key substitution as a design route to high-efficiency donors, with efficient hole transferterm·key at a small HOMOterm·dom offset

Full text — tagged

The reference list is not part of the text we hold for this paper. Our copy was extracted from the publisher's page and the bibliography did not survive that extraction, so the article below ends before its references. Nothing has been removed by us.

11.4% Efficiency non-fullereneterm·key polymer solar cellsterm·dom with trialkylsilyl substituted 2D-conjugated polymer as donor Abstract Simutaneously high open circuit voltageterm·key and high short circuit current density is a big challenge for achieving high efficiency polymer solar cellsterm·dom due to the excitonic nature of organic semdonductors. Herein, we developed a trialkylsilyl substituted 2D-conjugated polymer with the highest occupied molecular orbitalterm·dom level down-shifted by Si–C bond interaction. The polymer solar cellsterm·dom obtained by pairing this polymer with a non-fullereneterm·key acceptor demonstrated a high power conversion efficiencyterm·mth of 11.41% with both high open circuit voltageterm·key of 0.94 V and high short circuit current density of 17.32 mA cm −2 benefitted from the complementary absorption of the donor and acceptor, and the high hole transferterm·key efficiency from acceptor to donor although the highest occupied molecular orbitalterm·dom level difference between the donor and acceptor is only 0.11 eV. The results indicate that the alkylsilylterm·key substitution is an effective way in designing high performance conjugated polymer photovoltaic materials.

Bulk-heterojunction polymer solar cellsterm·dom (PSCsterm·dom) are composed of a blend active layer of a p -type conjugated polymer as donor and an n -type semiconductor as acceptor sandwiched between an anode and a cathode where at least one of the two electrodes should be transparent 1 2 3 . The n -type semiconductors used as acceptor in the PSCsterm·dom include fullerene derivatives 4 , n -type conjugated polymers 5 6 7 8 9 10 11 12 and n -type organic semiconductors ( n -OS) 13 14 15 16 17 18 19 20 21 22 23 . Among the acceptors, n -OS acceptors show distinguished advantages of easy tuning of absorption and electronic energy levels, strong absorbance, good morphology stability and more suitable for flexible devices. Therefore, the non-fullereneterm·key PSCsterm·dom with n -OS as acceptor have attracted great attention recently, and their power conversion efficiencyterm·mth (PCEterm·mth) has rapidly increased to 9–11% (refs 24 , 25 , 26 ).

In PSCsterm·dom, due to the low dielectric permittivity and relatively large excitons binding energy of organic semdonductor, polymer donor and acceptor with cascading enegy levels (the lowest unoccupied molecular orbital (LUMO) and highest occupied molecular orbitalterm·dom (HOMOterm·dom) levels of the conjugated polymer donor should be higher than the correponding LUMO and HOMOterm·dom levels of the acceptor) are specially required to provide a driving force for excitons dissociation at the heterojunction interface 27 28 29 30 . In this cascade model, it is generally accepted that the LUMO/HOMO energy levelsterm·dom differences between the donor and acceptor (Δ E LUMO and Δ E HOMOterm·dom ) should be larger than 0.3 eV for efficient excitons dissociation to overcome binding energy (usually 0.3–0.5 eV) of the excitons produced in the donor and acceptor phases by absorbing photons 27 30 31 32 . One crucial issue of the PSCsterm·dom is their relatively large photon energy lossterm·mth ( E loss ) at 1 sun illuminaiton which is defined as E loss = E g –e V oc , where E g is the lowest optical bandgap of the donor and acceptor components 27 33 34 35 . The reported E loss in the most efficient fullerene-based PSCsterm·dom is typically 0.7–1.0 eV. The larger E loss is associated with the larger driving force for excitons dissociation and the relative large non-radiative recombinations 27 28 36 37 , thus creating a great challenge to simultaneously obtain a large open circuit voltageterm·key ( V oc ) and a high short circuit current density ( J sc ) in the PSCsterm·dom.

In the non-fullereneterm·key PSCsterm·dom, medium bandgap polymer donor–low bandgap n-OS acceptor pairs are particular interesting due to their complementary absorption in the visible-near infrared region, providing an effective approch for getting a high J sc . In these systems, another interesting feature is the easy energy level matching to get higher V oc with lower E loss of 0.6–0.7 eV (refs 21 , 25 , 26 , 38 ). For example, the non-fullereneterm·key PSCsterm·dom based on a medium bandgap D-A copolymer PffT2-FTAZ-2DT as donor and a low bandgap n-OS IEIC (ref. 19 ) ( E g of 1.59 eV) as acceptor with a Δ E HOMOterm·dom of 0.17 eV showed a PCEterm·mth of 7.3% with a V oc of 0.998 V and a E loss of ca. 0.60 eV (ref. 38 ). The device based on a 2D-conjugated D-A copolymer PBDB-T as donor and a low bandgap n -OS ITICterm·key (ref. 20 ) ( E g of 1.59 eV) as acceptor with a Δ E HOMOterm·dom of 0.18 eV demonstrated a PCEterm·mth of 11.21% with a V oc of 0.899 V and a E loss of 0.64 eV (ref. 26 ). Notably, the lower E loss was attained with a relatively low Δ E HOMOterm·dom , suggesting a low driving force needed for excitons dissociation in this system. These encouraging results provide a plausible approach to well remove the trade-off between the V oc and J sc . Recently, we developed a series medium bandgap 2D-conjugated D-A copolymers based on bithienyl-benzodithiophene (BDTT) donor unit and fluorine-substituted benzotriazole (FBTA) acceptor unit 25 39 , which showed good photovoltaic performance in the PSCsterm·dom with the polymers as donor and n-OS as acceptors 11 25 . Among the polymers, J61 as donor blending with the n -OS ITICterm·key as acceptor with a Δ E HOMOterm·dom of 0.16 eV in the non-fullereneterm·key PSCsterm·dom displayed a high PCEterm·mth of 9.53% with a V oc of 0.89 V and a E loss of 0.68 eV (ref. 25 ). The high photovoltaic performance of the non-fullereneterm·key PSCsterm·dom with the small Δ E HOMOterm·dom between donor and acceptor is a very important phenomenon for the development of high performance PSCsterm·dom by increasing V oc and decreasing E loss .

To probe the possiblity of further increasing the V oc and PCEterm·mth of the non-fullereneterm·key PSCsterm·dom with smaller Δ E HOMOterm·dom value, herein we try to further decrease HOMO energy levelterm·dom of the BDTT-FBTA-based 2D-conjugated D-A copolymers by introducing trialkylsilyl substituents on BDTT unit. The trialkylsilyl substituents instead of alkyl side chains make HOMO energy levelterm·dom of the synthesized polymer J71 down-shifted to −5.40 eV and its absorbance enhanced in comparison with its analogue polymer J52 with alkyl side chains 25 . Very interestingly, the non-fullereneterm·key PSCterm·dom based on J71 as donor and ITICterm·key as acceptor with a smal Δ E HOMOterm·dom of 0.11 eV demonstrates a high PCEterm·mth of 11.41% with high V oc of 0.94 V, high J sc of 17.32 mA cm −2 and a low E loss of 0.63–0.65 eV. The PCEterm·mth of 11.41% is one of the highest efficiency of the single junction PSCsterm·dom reported in literatures till now. The results indicate that the alkylsilylterm·key substitution is an effective way in designing high performance conjugated polymer photovoltaic materials.

Results Design and synthesis of J71 In the molecular design of conjugated polymer donor materials for PSCsterm·dom, the main chain engineering (such as alternative donor (D)-acceptor (A) copolymerization) 40 41 and side chain engineering (such as conjugated side chains and electron-withdrawing substituents) 30 42 are widely used strategy to tune absorption spectra and electronic energy levels of the conjugated polymers. Silicon (Si)-containing fused rings, such as dibenzosilole and dithienosilole, are representative donor units in the high performance D-A copolymer donor materials. Silicon bridging atom in the Si-containing fused rings plays an important role in improving the planarity of the backbone and modifying the electronic properties of the polymers 43 44 45 46 . The effect of the Si-atomterm·dom on the electronic properties are mostly related to the bond interaction of the low-lying σ * orbital of the Si atomterm·dom with the π * orbital of the aromatic units, which results in stabilization of the LUMO energy level and lowering of HOMO energy levelterm·dom 47 . For example, Si-bridged bithiophene (dithienosilole, DTS) has lower HOMOterm·dom and LUMO energy levels compared with those of its C-bridged counterpart (CDT) 47 . In addition, the Si-bridging atom also showed a significant effect on the crystallinity of the polymer due to the longer C–Si bond length compared with the C–C bond, leading to stronger π – π stacking. Inspired by the special-function of the Si-atomterm·dom connecting with conjugated system, herein we designed another BDTT-FBTA-based D-A copolymer J71 by introducing alkylsilylterm·key substituents on thiophene conjugated side chains of BDTT unit for down-shifting HOMO energy levelterm·dom and strengthening interchain interaction of the polymers by the σ * (Si)– π *(C) bond interaction.

The polymer J71 was synthesized by Stille-coupling polycondensation between M1 (BDTT-Si) and M2 (FTAZ-based mononmer) 48 , as shown in Fig. 1a . The structure of BDTT-Si obtained by X-ray crystallography was also shown in Fig. 1a , and the crystallographic data of BDTT-Si (CCDC number: 1478875) are listed in Supplementary Table 1 and Supplementary Table 2 .

To better understand the effect of alkylsilylterm·key substitution on the electronic properties of the monomer and polymer, we compared the absorption spectra and electronic energy levels of BDTT-Si with the alkylsilylterm·key substituents and BDTT-C with alkyl substituents on the thiophene conjugated side chains of the BDTT units, as shown in Fig. 1b . It can be seen that the absorption maximum ( λ max ) of BDTT-Si is at 380 nm which is 5 nm red-shifted than that (375 nm) of BDTT-C. And the absorption coefficient of BDTT-Si is higher than that of BDTT-C, indicating that alkylsilylterm·key substitution affords not only broader but also stronger absorption. HOMO energy levelsterm·dom of the two monomers were measured by cyclic voltammetry. The onset oxidation potentials ( ϕ ox ) of BDTT-Si and BDTT-C are 0.99 and 0.90 V versus Ag/AgCl ( Fig. 1c ), and the HOMO energy levelsterm·dom ( E HOMOterm·dom ) were calculated to be −5.35 eV for BDTT-Si and −5.26 eV for BDTT-C (the calculation equation will be discussed below in the section on the electronic energy level measurements of J71 ). The LUMO levels of BDTT-Si and BDTT-C, estimated from their absorption edges and HOMOterm·dom levels, are −3.40 and −3.27 eV respectively. Clearly, alkylsilylterm·key substituted BDTT-Si has lower HOMOterm·dom and LUMO enrgy levels compared with those of its alkyl substituted counterpart BDTT-C ( Fig. 1d ). The results suggest that the alkylsilylterm·key side chain engineering can afford the similar effect of the Si–C bond interaction as that in Si-bridged aromatic compounds 47 , while the alkylsilylterm·key side chain approach developed in this work is more simple and convenient in down-shifting the HOMO energy levelterm·dom and strengthening the absorption. To further understand the effect of the alkylsilylterm·key side chains on the electronic energy levels of the monomers, we performed the theoretical calculation by the DFT method on the molecules at the DFT B3LYP/6-31G(d) level with the Gaussian 03 program package. The calculated HOMOterm·dom and LUMO energy levels of BDTT-Si are lower than those of BDTT-C ( Supplementary Fig. 1 ), which is consistent with the experimental results mentioned above.

The synthesis details of J71 were described in the Method section. J71 exhibits good solubility in many chlorinated solvents, such as chloroform, chlorobenzene and dichlorobenzene. The number average molecular weight ( M n ) of J71 is 23.5 kDa with a polydispersity index of 2.0, which was measured by high temperature Gel permeation chromatography (GPC). Thermogravimetric analysis (TGA) demonstrated a good thermal stability of the polymer with a 5% weight-loss temperature at 354 °C, as shown in Supplementary Fig. 2 .

Absorption spectra and electronic energy levels Figure 2a shows the molecular structures of the polymer J71 and the n-OS ITICterm·key acceptor, and Fig. 2b displays the absorption spectra of J71 solution and film together with the film absorpiton of ITICterm·key for comparison. In solution, J71 shows defined absorption profile with two peaks at about 528 and 573 nm, which can be ascribed to the vibronic bands associated with the (0-1) and (0-0) transitions respectively. The emergence of these features is believed to be associated with the partial aggregation of the fluorinated polymer chains. The absorption of J71 film is red-shifted by about 14 nm along with a stronger (0-0) transition peak in the long wavelength range. The red-shift of its absorption in the solid film indicates more ordered structure and stronger π–π stacking interaction, which should benefit higher hole mobility and better photovoltaic performance of the polymer. The absorption edge of J71 film is at 632 nm, which correponds to an optical bandgap of 1.96 eV. The film maximum extinction coefficient of J71 is 0.96 × 10 5 cm −1 , which is higher than its polymer analogue J52 with alkyl chain (0.73 × 10 5 cm −1 ) 25 . In addition, as one of the series of BDTT-FBTA-based copolymers 11 25 39 , J71 also demonstrates well complementary absorption with that of ITICterm·key n-OS acceptor in the wavelength range of 400–800 nm, which is beneficial for light harvesting in the non-fullereneterm·key PSCsterm·dom.

The HOMOterm·dom and LUMO energy levels of J71 were measured by electrochemical cyclic voltammetry 49 50 , for investigating the effect of the tripropylsilyl substitution on the electronic energy levels of the 2D-conjugated polymers. The HOMOterm·dom/LUMO energy levels ( E HOMOterm·dom / E LUMO ) can be calculated from onset oxidation/reduction potentials ( ϕ ox / ϕ red ) in the cyclic voltammograms according to the equations of E HOMOterm·dom / E LUMO =−e ( ϕ ox / ϕ red +4.8− ϕ Fc/Fc + ) (eV) 49 50 where ϕ Fc/Fc + is the redox potential of ferrocene/ferrocenium (Fc/Fc + ) couple in the electrochemical measurement system, and the evergy level of Fc/Fc + was taken as 4.8 eV below vacuum. The electrochemical measurement was performed in a 0.1 M acetonitrile solution of tetrabutylammonium hexafluorophosphate (n-Bu 4 NPF 6 ) with the sample ( J71 or ITICterm·key) film deposited on Pt disc electrode as working electrode, Ag/AgCl as reference electrode. ϕ Fc/Fc + was measured to be 0.44 V versus Ag/AgCl in this measurement system, and then the calculation equations are E HOMOterm·dom / E LUMO =−e ( ϕ ox / ϕ red +4.36) (eV).

Figure 2c shows the cyclic voltammegram of J71 film, from which the onset oxidation potential ( ϕ ox ) and onset reduction potential ( ϕ red ) of J71 were measured to be 1.04 and −1.10 V versus Ag/AgCl respectively ( Supplementary Fig. 3 ). The HOMO energy levelterm·dom ( E HOMOterm·dom ) and LUMO energy level ( E LUMO ) of J71 were calculated to be −5.40 and −3.24 eV respectively (see Fig. 2d ), according to the Equations mentioned above. Under the same experimental conditions, the E HOMOterm·dom and E LUMO of ITICterm·key were measured to be −5.51 and −3.84 eV respectively (see Fig. 2d and Supplementary Fig. 3b ).

It should be noted that the E HOMOterm·dom of J71 is 0.19 and 0.08 eV deeper than that of its polymer analogue J52 with alkyl substituent ( E HOMOterm·dom =−5.21 eV, see Supplementary Fig. 4 ) and J61 with alkylthio substituent ( E HOMOterm·dom =−5.32 eV) respectively 25 . The results indicate that the alkylsilylterm·key substitution (with σ * (Si)– π *(C) bond interaction) is effective in lowering the HOMO energy levelterm·dom of the 2D-conjugated polymers. It should be mentioned that the electrochemical bandgap ( E LUMO - E HOMOterm·dom ) of J71 is 2.16 eV, which is 0.2 eV larger than that (1.96 eV) of its optical bandgap. The larger electrochemical bandgap is a common phenomenon for the conjugated polymers and is reasonable in considering the energy barriers needed for the charge transfer in electrochemical oxidation and reduction reactions on the electrode.

Photovoltaic properties PSCsterm·dom were fabricated with a traditional device structure of ITO (indium tin oxide) /PEDOT: PSS (poly (3, 4-ethylenedioxythiophene): poly (styrene-sulfonate)) / J71 : ITICterm·key (1:1, w/w) /PDINO 51 (perylene diimide functionalized with amino N -oxide)/Al, where PDINO was chosen as the cathode interlayer for lowering the work funciton of Al 51 . The weight ratio of J71 donor: ITICterm·key acceptor is 1: 1 according to our recent work in the studies of non-fullereneterm·key PSCsterm·dom based on J61:ITICterm·key 25 , considering the similar chemical structure of J71 with J61. The active layers with a thickness of about 100 nm were prepared by spin-coating the J71 : ITICterm·key blend solution with a total blend concentration of 12 mg ml −1 in chloroform at 3000, r.p.m. Figure 3a shows the current density–voltage ( J–V ) curves, and the corresponding incident photon to converted current efficiency (IPCE) spectra are shown in Fig. 3b . For a clear comparison, the detailed photovoltaic performance data are listed in Table 1 . The as-prepared PSCsterm·dom based on J71 : ITICterm·key showed a PCEterm·mth of 9.03% with a high V oc of 0.96 V, a J sc of 14.81 mA cm −2 and a FF of 63.63%. After thermal annealingterm·mth at 150 °C for 10 min, the PCEterm·mth of the PSCterm·dom was further improved to 11.41% with a high V oc of 0.94 V, a high J sc of 17.32 mA cm −2 and a FF of 69.77%. The PCEterm·mth of 11.41% is one of the highest efficiency for the single junction PSCsterm·dom reported in literatures so far. And very importantly, both high V oc and high J sc were achieved in the J71 :ITIC-based PSCsterm·dom, which is a big challenge for the high performance PSCsterm·dom.

It can be seen that the PSCsterm·dom based on J71 : ITICterm·key exhibited high V oc in the range of 0.94–0.96 V, which is certainly benefitted from the low-lying HOMO energy levelterm·dom of the alkylsilylterm·key substituted J71 . As mentioned in the Introduction part, a crucial issue in the studies of PSCsterm·dom is to minimize the device photon energy lossterm·mth ( E loss ) which is defined as E loss = E g –e V oc , where E g is the lowest optical band gap among the donor and acceptor components 33 34 35 . In the present PSCsterm·dom based on J71 : ITICterm·key, the lowest E g is 1.59 eV for the ITICterm·key acceptor with onset absorption at 782 nm (see Supplementary Fig. 5b ). Therefore the V oc of 0.94–0.96 V results in a low E loss of 0.63–0.65 eV, which is smaller than that of most PSCsterm·dom and approaching the empirically low threshold of 0.6 eV. It should be mentioned that using the onset absorption to determine E g is the commonly accepted and wide used method by different groups 33 34 35 52 53 . And this method can provide a straight forward comparison of our results with those results previously reported (see Supplementary Table 3 ). Recently, Scharber et al . proposed a more accurate method to measure the E g value of the active layer (blend of donor and acceptor) of the PSCsterm·dom from IPCE spectrum 36 . With this method, we obtained a E g of 1.58 eV from the onset of IPCE spectrum of the PSCterm·dom based on J71 :ITICterm·key (as shown in Supplementary Fig. 6 ) which is consistent with the E g value of 1.59 eV from the absorption edge of ITICterm·key mentioned above.

The plot of e V OC against E g , and the plots of PCEterm·mth and IPCE max against E loss for the PSCsterm·dom reported in literatures are provided in Supplementary Fig. 7 , and the detailed data of the corresponding devices are listed in Supplementary Table 3 for a clear comparison. It should be noted that the PCEterm·mth and the IPCE max (76.5%) of the PSCterm·dom based on J71/ ITICterm·key are in fact the highest values among the fullerene and non-fullereneterm·key PSCsterm·dom reported in literatures with E loss less than 0.65 eV. Such small E loss is benefitted from the small Δ E HOMOterm·dom of 0.11 eV in the PSCsterm·dom based on J71/ ITICterm·key.

The IPCE spectra of the PSCsterm·dom based on J71 : ITICterm·key ( Fig. 3b ) demonstrate broad and high photo-response from 300 to 790 nm, which indicates high photo-conversion efficiency for the absorptions of both J71 polymer donor and ITICterm·key acceptor. The IPCE spectra in the wavelength range of 650–790 nm are corresponding to the excitons dissociation of the ITICterm·key acceptor with the hole transferterm·key from the HOMOterm·dom of ITICterm·key to that of J71 . The high IPCE values in this range confirm that efficient hole transferterm·key process did occour even though the Δ E HOMOterm·dom between J71 donor and ITICterm·key acceptor is only 0.11 eV. The J sc values integrated from the IPCE spectra are 16.564 mA cm −2 for the thermal annealed device and 14.688 mA cm −2 for the as-prepared device, which agrees well with those values obtained from the J–V curves within 5% mismatch ( Table 1 ). It should be noted that the low Δ E HOMOterm·dom request can provide big chance in the molecular design of photovoltaic materials, such as for the polymer donor to further downshift its E HOMOterm·dom toward a larger V oc , thus it will be promising to address the big challenge of PSCsterm·dom for maximizing V oc and J sc at the same time and ultimately getting a high-efficiency.

We also tried to fabricate the inverted PSCsterm·dom with a device structure of ITO/ZnO / J71 : ITICterm·key (1:1, w/w) /MoO 3 /Al. The inverted PSCterm·dom with thermal annealingterm·mth at 150 °C for 10 min showed a PCEterm·mth of 10.7% with a V oc of 0.93 V, a J sc of 17.36 mA cm −2 and a FF of 66.05%, as shown in Supplementary Fig. 8 . The slightly lower PCEterm·mth of the inverted device could be due to the electrode buffer layer materials used in the inverted PSCsterm·dom, and the optimization of the inverted PSCsterm·dom is underway.

The charge carrier mobilities of the PSCsterm·dom were measured by space-charge-limited current (SCLC) method to investigate the effect of thermal annealingterm·mth. The plots of the current density versus voltage of the devices for the mobility measurements are shown in Supplementary Fig. 9 . For the as-cast blend film, their hole mobility μ h and electron mobility μ e are estimated to be 2.08 × 10 −4 cm 2 V −1 s −1 and 0.96 × 10 –4 cm 2 V −1 s −1 respectively with μ h / μ e of 2.17, while after thermal annealingterm·mth the μ h and μ e values were increased to 3.78 × 10 –4 cm 2 V −1 s −1 and 3.07 × 10 –4 cm 2 V −1 s −1 respectively with the improved μ h / μ e ratio of 1.23. The improved photovoltaic performance of the PSCsterm·dom with thermal annealingterm·mth could be ascribed to the higher and more-balanced charge carrier mobilities of the thermal-treated J71 : ITICterm·key film. Figure 3c shows the dependance of J sc values on the light-intensity ( P light ) which reflects the charge recombination behaviour of the devices. In general, the relationship between J sc and P light can be described as J sc ∝ P light α . If the bimolecular recombination of the charge carriers is negligible, the power-law exponent α should be unity. The values of α for the as-prepared and thermal treated devices are 0.922 and 0.986, respectively ( Fig. 3c ). The higher α value of the thermal-treated device suggests the reduced charge recombination in its device, which correlates well with the balanced charge carrier mobilities of the devices mentioned above.

The effect of thermal annealingterm·mth on the device performance was further studied by analysing the series resistance ( R s ) and shunt resistance ( R sh ) of the devices from their dark J–V curves ( Fig. 3d ), and the R s and R sh values are also listed in Table 1 . The R s and R sh values of the as-prepared device are 10.38 Ω cm 2 and 0.57 kΩ cm 2 respectively, while after thermal annealingterm·mth, R s was decreased to 1.13 Ω cm 2 and R sh was increased to 1.9 kΩ cm 2 , suggesting the better overall diode characteristics after the thermal annealingterm·mth, which is also reflected in the lower ideality factor n (1.51 for the thermal-annealed device in comparison with 1.92 for the as-prepared device) and lowest dark saturation current density J o (1.89 × 10 −10 mA cm −2 for the thermal-annealed device while it is 1.44 × 10 –8 mA cm −2 for the as-prepared device).

Morphological characterization For non-fullereneterm·key PSCsterm·dom, morphology can be a determining factor governing the devcie performance 54 55 . Here, the microstructural features and surface morphologies of the neat J71 and ITICterm·key films as well as their blend films with or wihtout thermal annealingterm·mth were investigated by grazing incident wide-angle X-ray diffraction (GIWAXS) plots 56 and tapping-mode atomic force microscopy (AFM). Figure 4 shows the plots and images of the GIWAXS measurements. Strong diffraction peaks of the neat J71 film, as shown in Fig. 4b , reveal the semicrystalline structure and strong preference for face-on orientation in the polymer film. In addition, characteristic of the long range order and crystallinity in the film was observed in the in-plane direction. The high crystallinity is largely benefitted from its proper side chain 42 and fluorination effect 48 . For neat ITICterm·key film, its lamellar (100) peak is located at 0.360 Å −1 and π – π stacking (010) peak is at 1.75 Å −1 ( Fig. 4a,e ), corresponding to lamellar distance of 17.44 Å and a π – π stacking distance of 3.58 Å. Its large azimuthal distribution of the diffraction peaks suggests randomly oriented crystallites, which is related to the steric effect of its tetrahexylphenyl substituents.

For the blend films, the GIWAXS plots demonstrated microstructural features with the diffraction patterns contributed from individual components ( Fig. 4c,g ). After thermal annealingterm·mth, the J71 : ITICterm·key blend film is more prone to adopt a prominent face-on orientation ( Fig. 4d,h ) as evidenced by the small azimuthal distribution of the (010) π-π stacking in the out-of-plane direction. Further looking into diffraction patterns, it can be found that, the peak intensities become significantly stronger accompanied by their narrower peak widths ( Fig. 4d,h ). Furthermore, a 3.85 Å π-π -stacking distance ( q =1.63 Å −1 ) is seen in the thermal treated blend films, which is slighltly smaller than that of the as-cast J71 : ITICterm·key film (3.88 Å, q =1.62 Å −1 ). All the behaviour is related to the higher crystalline characteristics of the thermal treated blend film, and the combination of the preferred face-on orientation and the tight π-π -stacking of the polymer backbone is known to assist intermolecular charge transport and suppress the charge carrier recombination, which eventually improves the photovoltaic performance.

The AFM images, as shown in Supplementary Fig. 10 , reveal that the blend films have relatively smooth surface with a root-mean-square (RMS) roughness of 0.722 nm for the as-cast J71 : ITICterm·key blend film and 0.741 nm for the thermal annealed blend film, indicating the good miscibility between the alkylsilylterm·key substituted J71 polymer donor and ITICterm·key acceptor. From the AFM phase images in Supplementary Fig. 10d , it can be seen that with the thermal annealingterm·mth, enhanced crystalline domains with nano-networks around 20 nm are visualized. This highly intercrystalline morphology asross the active layer enhances the polymer donor domain connectivity and thereby improves the hole transport, which correlates well with its superior device performance.

Hole transferterm·key and charge separation dynamics As mentioned above, it is likely that hole transferterm·key from ITICterm·key to J71 is highly efficient, despite the energy difference between HOMOterm·dom levels (Δ E HOMOterm·dom ) of ITICterm·key and J71 is only 0.11 eV which is much smaller than the empirical threshold of 0.3 eV for effective exciton dissociation. To confirm the assessment, we performed transient absorptionterm·key spectroscopy measurement to investigate the charge transfer dynamics of photo-induced carriers in the blend film of J71 : ITICterm·key. The primary absorption peaks for J71 and ITICterm·key are well separated in spectral domain ( Fig. 2b ), so we can extract the spectral and temporal characteristics of hole transferterm·key dynamics with selected excitation. The pump wavelengths of 540 and 710 nm were selected to excite J71 and ITICterm·key, respectively. The excitation density is kept in a weak regime below 1 μJ cm −2 to avoid the effect of exciton-exciton annihilation. The pump-probe experimentsterm·mth measure the pump-induced differential change of the probe transmission, Δ T / T =( T pump-on − T pump-off )/ T pump-off .

Figure 5 shows the results of transient absorptionterm·key spectroscopy with excitation wavelength at 710 nm. At this pump wavelength, only ITICterm·key is excited as confirmed by the absence of transient absorptionterm·key signal from the neat sample of J71 ( Supplementary Fig. 11 ). A bleaching signal peaked at 710 nm appears in both neat ITICterm·key and the blend ( Fig. 5a,b ). This signal at the absorption peak of ITICterm·key can be naturally ascribed to the ground state bleaching (GSB) of the transition in ITICterm·key. In addition, clear bleaching peaks at 540 and 590 nm appear in the transient absorptionterm·key spectrum of blend ( Fig. 5b ). These wavelengths are exactly the absorption peaks of J71 ( Fig. 2b ). The spectral feature is also consistent with the GSB signal from J71 observed with resonant excitation at 540 nm ( Supplementary Fig. 11 ). Figure 5c compares the transient absorptionterm·key spectra from the blend sample at different time delays with the initial GSB spectra observed in J71 and ITICterm·key under resonant excitations. Basically, the bleaching signals at 540 and 590 nm are built up with the decays of bleaching signal at 710 nm, suggesting the transfer of excitations from ITICterm·key to J71 . The excitation photon energy (at 710 nm) is much smaller than that required for exciton absorption of the polymer, therefore, the bleaching signals (540 and 590 nm) cannot be ascribed to energy transfer process. Notably, the bleaching signals at about 540 nm and about 590 nm appear simultaneously with the decay of the GSB signals of ITICterm·key at about 710 nm ( Fig. 5c ), such excitation transfer can be naturally assigned to the hole transferterm·key since the LUMO level of ITICterm·key is much lower than that of J71 .

Figure 5d compares the dynamics probed at 710 nm for the films of neat ITICterm·key and blend J71 :ITICterm·key in a normalized scale. For the early stage at less than 20 ps, the relaxation rate becomes dramatically faster in the blend film with respect to the neat ITICterm·key, indicating the presence of additional relaxation channel of hole transferterm·key in the blend film. We quantify the early-stage kinetics with a bi-exponetial decay function ( Supplementary Fig. 12 ). The lifetime parameters for the two components are about 0.72 ps and about 15.0 ps respectively for the neat ITICterm·key which decreases to about 0.29 ps and about 4.5 ps in the blends. Correspondingly, the buildup of signals probed at 540 and 590 nm in blend film is different from the abrupt rises observed in J71 excited at 570 nm ( Supplementary Fig. 11 ). The onset shows two exponential growth components with lifetime parameters of about 0.3 ps and about 4.8 ps, respectively ( Fig. 5e ), confirming the hole transferterm·key is the primary origin of the GSB signal at the absorption band of J71 . With these results, we briefly summarize the dynamics in Fig. 5f where the hole transferterm·key from ITICterm·key is quite efficient with rate up to 3 ps −1 . Moreover, it is worthy noting that the GSB signals at 540, 590 and 710 nm in the blend film persist to nanosecond scale which is much longer than those in neat samples of J71 and ITICterm·key. The results suggest the existence of the long-lived dissociated excitons in the blend which is beneficial for electricity generation.

Discussion In conclusion, another BDTT-FBTA-based 2D-conjugated D-A copolymer J71 with tripropylsilyl substituents on thiophene conjugated side-chain of BDTT unit, was synthesized. J71 possesses a low-lying HOMO energy levelterm·dom at −5.40 eV and high film absorption extinction coefficients which is benefitted from the σ * (Si)– π *(C) bond interaction with the trialkylsilyl substitution. Non-fullereneterm·key PSCsterm·dom were fabricated with J71 as donor and a narrow bandgap n-OS ITICterm·key as acceptor, and the PCEterm·mth of the PSCsterm·dom based on J71 /ITICterm·key (1:1, w/w) with thermal annealingterm·mth at 150 °C for 10 min reached 11.41% with simultaneously high V oc of 0.94 V and a high J sc of 17.32 mA cm −2 . The PCEterm·mth of 11.41% is one of the highest efficiencies of the single-junction PSCsterm·dom reported in literatures till now. The high V oc , which results in a low E loss of 0.63–0.65 eV for the PSCterm·dom, is benefitted from the low-lying HOMOterm·dom level of the J71 donor. The high J sc should be ascribed to the extraordinarily high exciton charge separation efficiency even though the HOMOterm·dom energy difference between the donor and acceptor is quite small (only 0.11 eV). And the high hole transferterm·key efficiency from ITICterm·key to J71 was confirmed by transient absorptionterm·key spectra. The results indicate that the alkylsilylterm·key substitution is an effective way in designing high performance conjugated polymer photovoltaic materials and the driving force for the hole transferterm·key from the acceptor to donor could be quite small in the non-fullereneterm·key PSCsterm·dom with the n-OS as acceptor.

Methods Materials and synthesis ITICterm·key was purchased from Solarmer Materials Inc. The other chemicals and solvents were purchased from J&K, Alfa Aesar or TCI Chemical Co. The monomers and polymer J71 were synthesized according to Scheme 1. The Benzo[1,2-b:4,5-b′]dithiophene-4,8-dione and M2 was synthesized according to the procedure reported in the literatures 39 .

Monomer 1 was synthesized as follows: Under protection of argon, to a solution of thiophene (8.4 g, 100 mmol) in THF (100 ml) at −78 °C was added n -BuLi (40 ml, 2.5 M in hexane) slowly, the mixture was kept at −78 °C for 1 h and warmed slowly to room temperature. Then, chlorotripropylsilane (19.2 g, 100 mmol) was added, and the mixture stirred overnight. The mixture was extracted by diethyl ether twice, washed by water and brine. Further purification was carried out by column chromatography using hexane as eluent to obtain pure tripropyl(thiophen-2-yl)silane ( 1 ). as a colourless liquid. (22.3 g, 93% yield). 1 H NMR(400 MHz, CDCl 3 ), δ (p.p.m.): 7.57-7.56 (d, 1H), 7.22 (d, 1H), 7.16-7.15(t, 1H), 1.40-1.32 (m, 6H), 0.96-0.92(m, 9H), 0.80-0.75(m, 6H).

Monomer BDTT-Si was synthesized as follows: Under protection of argon, to a solution of compound 1 (7.2 g, 30 mmol) in THF (30 ml) at 0 °C was added n -BuLi (12 ml, 2.5 M in hexane), the mixture was kept at 0 °C for 15 min and heated to 50 °C and strirred for 2 h. Then, Benzo[1,2-b:4,5-b′]dithiophene-4,8-dione (2.2 g, 10 mmol) was added quickly, and the mixture stirred for 2 h. After cooling down to the room temperature, SnCl 2 ·2H 2 O (18 g, 80 mmol) in 10% HCl (35 ml) was added, and the mixture was stirred for 3 h. The mixture was extracted by diethyl ether twice, washed by water and brine. The crude product purified with column chromatography using petroleum ether as eluent to obtain pure 4,8-bis(5-(tripropylsilyl)thiophen-2-yl)benzo[1,2-b:4,5-b′]dithiophene ( BDTT-Si ) as a light yellow solid. (2.8 g, 42% yield). 1 H NMR (400 MHz, CDCl 3 ), δ (p.p.m.): 7.65-7.64(d, 2H), 7.57-7.56 (d, 2H), 7.47-7.45 (d, 2H), 7.36-7.35(d, 2H), 1.53-1.42(m, 12H), 1.05-1.00(m, 18H), 0.92-0.86 (m, 12H). 13 C NMR (100 MHz, CDCl 3 ), δ (p.p.m.):145.0, 139.3, 139.1, 136.6, 134.9, 129.3, 127.7, 124.2, 123.6, 18.6, 17.6, 16.5.

Monomer M1 was synthesized as follows: To a solution of BDTT-Si (1.332 g, 2 mmol) in THF (20 ml) at −78 °C was added n -BuLi (2 ml, 2.5 M in hexane). After the addition, the mixture was kept at −78 °C for 40 min; trimethyltin chloride (6 ml, 1 M in THF) was added dropwise. The resulting mixture was stirred for 2 h at room temperature. Then, it was poured into water and extracted with diethyl ether, washed by water and brine and after drying over MgSO 4 , the solvent was removed and the residue was recrystallized with methanol to afford a yellow crystal (5,5′-(2,6-bis(trimethylstannyl)benzo(1,2-b:4,5-b′)dithiophene-4,8-diyl)bis(thiophene-5,2-diyl))bis(tripropylsilane) ( M1 ) (1.56 g, 78.8% yield). 1 H NMR (400 MHz, CDCl 3 ), δ (p.p.m.): 7.74-7.67(t, 2H), 7.60–7.59(d, 2H), 7.36(d, 2H), 1.53–1.45(m, 12H), 1.04–0.97(m, 18H), 0.91–0.87(m, 12H), 0.46-0.32(t, 18H). 13 C NMR (100 MHz, CDCl 3 ), δ (p.p.m.):205.9, 144.9, 142.3, 141.4, 137.8, 136.5, 133.9, 130.4, 128.0, 121.5, 30.0, 17.6, 16.6, 15.3, −7.4, −7.5, −9.3, −11.0, −11.1.

Polymer J71 was synthesized as follows: M1 (248 mg, 0.25 mmol) and M2 (175 mg, 0.25 mmol) and dry toluene (10 ml) were added to a 25 ml double-neck round-bottom flask. The reaction container was purged with argon for 20 min, and then Pd(PPh 3 ) 4 (10 mg) was added. After another flushing with argon for 20 min, the reactant was heated to reflux for 12 h. The reactant was cooled down to room temperature and poured into MeOH (200 ml), then filtered through a Soxhlet thimble, which was then subjected to Soxhlet extraction with methanol, hexane, and chloroform. The polymer J71 of 275 mg (Yield 91%) was recovered as solid from the chloroform fraction by precipitation from methanol, and was dried under vacuum. GPC: M n =23.5 kDa; M w /M n =2.0. Anal. Calcd for C 66 H 85 F 2 N 3 S 6 Si 2 (%): C, 65.68; H, 7.10; N, 3.48. Found (%):C, 64.79; H, 7.06; N, 3.55. 1 H NMR (CDCl 3 , 400 MHz): δ (p.p.m.) 8.15–8.12 (br, 2H), 8.01–7.45 (br, 6H), 7.20–6.90 (br, 2H), 4.69 (br, 2H), 2.24–0.83 (br, 73H).

General characterization 1 H NMR spectra were measured on a Bruker DMX–400 spectrometer with d –chloroform as the solvent and trimethylsilane as the internal reference. Ultraviolet–visible absorption spectra were measured on a Hitachi U–3010 Ultraviolet–vis spectrophotometer. Mass spectra were recorded on a Shimadzu spectrometer. Elemental analyses were carried out on a flash EA 1112 elemental analyser. Thermogravimetric analysis (TGA) was conducted on a Perkin–Elmer TGA–7 thermogravimetric analyser at a heating rate of 20 °C min −1 under a nitrogen flow rate of 100 ml min −1 . Gel permeation chromatography (GPC) measurements was performed on Agilent PL-GPC 220 instrument with high temperature chromatograph, using 1,2,4-trichlorobenzene as the eluent at 160 °C. Photoluminescience (PL) spectra were measured with a Shimadzu RF-5301PC fluorescence spectrophotometer. Electrochemical cyclic voltammetry was performed on a Zahner IM6e Electrochemical Workstation under a nitrogen atmosphere. The cyclic voltammograms of J71 polymer film and ITICterm·key film on a Pt disk electrode (working electrode) were measured with a potential scan rate of 20 mV s −1 in an acetonitrile solution of 0.1 M tetrabutylammonium hexafluorophosphate (n-Bu 4 NPF 6 ) with a Ag/AgCl reference electrode and a platinum wire counter electrode. The ferrocene/ferrocenium (Fc/Fc + ) couple was used as an internal reference. Thin films of J71 or ITICterm·key were prepared by drop-casting 1.0 μl of their chloroform solutions (analytical reagent, 1 mg ml −1 ) onto the working electrode and then dried in the air. The J71 or ITICterm·key film and J71 :ITICterm·key blend film for morphology measurements were prepared by spin-coating the corresponding solution in chloroform with a concentration of 10 mg ml −1 on indium tin oxide (ITO) glass at 3,000 rpm for 30 s. The film morphology was measured using an atomic force microscope (AFM, SPA-400) using the tapping mode.

Hole and electron mobilities were measured using the the space charge limited current (SCLC) method, with the hole-only device of ITO/PEDOT:PSS/J71: ITICterm·key(1:1, w/w) /Au for hole mobility measurement and electron-only device of ITO/ZnO/J71: ITICterm·key (1: 1, w/w) /PDINO/Al for electron mobility measurement. The SCLC mobilities were calculated by MOTT-Gurney equation: 57 58 Where J is the current density, ɛ r is the relative dieletiric constant of active layer material usually 2–4 for organic semiconductor, herein we use a relative dielectric constant of 4, ɛ 0 is the permittivity of empty space, μ is the mobility of hole or electron and L is the thickness of the active layer, V is the internal voltage in the device, and V = V app – V bi , where V app is the voltage applied to the device, and V bi is the built-in voltage resulting from the relative work function difference between the two electrodes (in the hole-only and the electron-only devices, the V bi values are 0.2 and 0 V respectively).

Grazing-incidence wide-angle X-ray scattering (GIWAXS) measurements were conducted at PLS-II 9A USAXS beam line of the Pohang Accelerator Laboratory in Korea. X-rays coming from the in-vacuum undulator (IVU) were monochromated (wavelength λ =1.109 94 Å) using a double crystal monochromator and focused both horizontally and vertically (450 (H) × 60 (V) μm 2 in fwhm at sample position) using K-B type mirrors. The GIWAXS sample stage was equipped with a 7-axis motorized stage for the fine alignment of sample, and the incidence angle of X-ray beam was set to be 0.13–0.135° for polymer films, ITICterm·key film and their bend films. GIWAXS patterns were recorded with a 2D CCD detector (Rayonix SX165), and X-ray irradiation time was 6–9 s, dependent on the saturation level of the detector. Diffraction angles were calibrated using a sucrose standard (monoclinic, P21, a =10.8631 Å, b =8.7044 Å, c =7.7624 Å, β =102.938°), and the sample-to-detector distance was ∼231 mm.

For transient absorptionterm·key spectroscopy, pump pulses with tunable wavelength were from an optical parametric amplifier pumped by a regenerative amplifier (Libra, Coherent, 1 kHz, 90 fs). The carrier dynamics was probed by a broadband supercontinuum light source that was generated by focusing a small portion of the beam of the amplifier onto a sapphire plate. The chirp of the probe supercontinuum was corrected with error to be less than 100 fs over the whole spectral range. The transient absorptionterm·key signal was then analysed by a high speed charge-coupled device (S11071-1104, Hamamatsu) with a monochromater (Acton 2358, Princeton Instrument) at 1 kHz enabled by a custom-built board from Entwicklungsbuero Stresing. The angle between the polarizations of pump and probe beam was set at the magic angle.

Device fabrication and characterization The PSCsterm·dom were fabricated with a structure of ITO/PEDOT: PSS (40 nm)/active layer/PDINO/Al. A thin layer of PEDOT: PSS was deposited through spin-coating on precleaned ITO-coated glass from a PEDOT: PSS aqueous solution (Baytron P VP AI 4083 from H. C. Starck) at 2,000 rpm and dried subsequently at 150 °C for 15 min in air. Then the device was transferred to a nitrogen glove box, where the active blend layer of J71 and ITICterm·key was spin-coated from its chloroform solution onto the PEDOT: PSS layer under a spin-coating rate of 3,000 r.p.m. After spin-coating, the active layers were annealed at 150 °C for 10 min for the devices with thermal annealingterm·mth treatment. The thickness of the active layers is about 100 nm. Then methanol solution of PDINO at a concentration of 1.0 mg ml −1 was deposited atop the active layer at 3,000 r.p.m. for 30 s to afford a PDINO cathode buffer layer with thickness of about 10 nm. Finally, top Al electrode was deposited in vacuum onto the cathode buffer layer at a pressure of about 5.0 × 10 −5 Pa. The active area of the devices was 4.7 mm 2 . Optical microscope (Olympus BX51) was used to define the active area of the devices. The current density–voltage ( J–V ) characteristics of the PSCsterm·dom were measured in a nitrogen glovebox (O 2 <0.1 p.p.m., H 2 O<0.1 p.p.m.) on a computer-controlled Keithley 2450 Source-Measure Unit. The voltage scan is from −1.5 to 1.5 V with voltage step of 10 mV and delay time of 1 ms. Oriel Sol3A Class AAA Solar Simulator (model, Newport 94023A) with a 450W xenon lamp and an air mass (AM) 1.5 filter was used as the light source. The light intensity was calibrated to 100 mW cm −2 by a Newport Oriel 91150V reference cell. Masks made by laser beam cutting technology with a well-defined area of 2.2 mm 2 were used to define the effective area for accurate measurement of the photovoltaic performance. All the measurements with mask or without mask gave consistent results with relative errors within 0.5%. Actually, the measurement with mask give a slightly higher PCEterm·mth mainly due to its slightly higher FF. The PCEterm·mth results in the manuscript are from the measurement without mask. PCEterm·mth statistics were obtained using 30 individual devices fabricated under the same conditions. The input photon to converted current efficiency (IPCE) was measured by Solar Cell Spectral Response Measurement System QE-R3-011 (Enli Technology Co., Ltd., Taiwan). The light intensity at each wavelength was calibrated with a standard single-crystal Si photovoltaic cell.

Data availability The data that support the findings of this study are available from the corresponding author on request.

Additional information How to cite this article: BinpersName, H. et al . 11.4% Efficiency non-fullereneterm·key polymer solar cellsterm·dom with trialkylsilyl substituted 2D-conjugated polymer as donor. Nat. Commun. 7 , 13651 doi: 10.1038/ncomms13651 (2016).

Supplementary Material Supplementary Figures 1-12, Supplementary Tables 1-3, Supplementary References.