T. L.
Nguyen
a,
H.
Choi
b,
S.-J.
Ko
b,
M. A.
Uddin
a,
B.
Walker
b,
S.
Yum
a,
J.-E.
Jeong
a,
M. H.
Yun
b,
T. J.
Shin
c,
S.
Hwang
a,
J. Y.
Kim
*b and
H. Y.
Woo
*a
aDepartment of Nanofusion Engineering, Department of Cogno-Mechatronics Engineering, Pusan National University, Miryang 627-706, Korea. E-mail: hywoo@pusan.ac.kr
bSchool of Energy and Chemical Engineering, Ulsan National Institute of Science and Technology (UNIST), Ulsan 689-798, Korea. E-mail: jykim@unist.ac.kr
cPohang Accelerator Laboratory, San 31, Hyoja-dong, Pohang 790-784, Korea
First published on 19th June 2014
We report a series of semi-crystalline, low band gap (LBG) polymers and demonstrate the fabrication of highly efficient polymer solar cells (PSCs) in a thick single-cell architecture. The devices achieve a power conversion efficiency (PCE) of over 7% without any post-treatment (annealing, solvent additive, etc.) and outstanding long-term thermal stability for 200 h at 130 °C. These excellent characteristics are closely related to the molecular structures where intra- and/or intermolecular noncovalent hydrogen bonds and dipole–dipole interactions assure strong interchain interactions without losing solution processability. The semi-crystalline polymers form a well-distributed nano-fibrillar networked morphology with PC70BM with balanced hole and electron mobilities (a h/e mobility ratio of 1–2) and tight interchain packing (a π–π stacking distance of 3.57–3.59 Å) in the blend films. Furthermore, the device optimization with a processing additive and methanol treatment improves efficiencies up to 9.39% in a ∼300 nm thick conventional single-cell device structure. The thick active layer in the PPDT2FBT:PC70BM device attenuates incident light almost completely without damage in the fill factor (0.71–0.73), showing a high short-circuit current density of 15.7–16.3 mA cm−2. Notably, PPDT2FBT showed negligible changes in the carrier mobility even at ∼1 μm film thickness.
Broader contextPolymer solar cells (PSCs) have great potential as a promising candidate for clean and renewable energy sources. PSCs are attracting increasing attention from both academia and industry, due to the demand for flexible, portable and solution processable low-cost photovoltaic devices to comply with increasing global energy demand. As results of many efforts for developing PSCs, the power conversion efficiency (PCE) of PSCs has been tremendously improved over the past decades. However, the low PCE is still a hurdle to overcome for commercial applications of PSCs. To further optimize PSCs, several challenges have to be considered including development of highly soluble and crystalline photovoltaic polymers to enable thick film production without damage in the fill factor. In this work, we report a new series of semi-crystalline photovoltaic polymers, demonstrating a clear molecular design strategy, structure–property relationships and the highest PCE of 9.39% (reported so far) in a 290 nm thick single-cell device. We believe that our findings offer a bright future for the large scale commercialization of PSCs into daily-use electronic devices. |
To further improve PCEs, first and foremost, the molecular structure of LBG polymers should be carefully designed by considering their close relationship with the photovoltaic parameters, including short-circuit current density (JSC), open-circuit voltage (VOC) and fill factor (FF). Recently, highly efficient photovoltaic materials have been designed by introducing fluorine (F) atoms onto the polymeric chain.10,11 Fluorine has a small van der Waals radius (∼1.35 Å) and is the most electronegative element with a Pauling electronegativity of 4.0. The introduction of fluorine onto the periphery of an electron-deficient unit is a versatile strategy, because it not only minimizes any undesired steric hindrance along the polymer chains but also effectively stabilizes the highest occupied molecular orbital (HOMO) and lowest unoccupied molecular orbital (LUMO) levels. Furthermore, hole mobility improves upon fluorination, though a reverse trend has been observed in some cases.12 The fluorine substituent often has a great influence on inter- and intramolecular interactions,13–15 which play important roles in the solid-state polymer organization with a cofacial π–π stacking. An effective approach to planarize a polymer chain (without losing its solution processability) is to create a noncovalent attractive interaction between neighboring moieties via intramolecular hydrogen bonds, dipole–dipole interactions, etc. Noncovalent intramolecular O⋯S interactions between alkoxy substituents and thiophene rings have been demonstrated to be effective for minimizing torsional angles within polymer backbones.16–18 On increasing the coplanarity of a polymer chain with close solid-state π–π stacking, both polaron and exciton delocalization and their transport characteristics can be improved.19–21
This study reports highly efficient new LBG polymer structures with a planar polymeric backbone formed via noncovalent intra- and interchain hydrogen bonds and dipole–dipole interactions, leading to highly ordered film morphologies, deep HOMO level, balanced electron and hole mobilities (a hole/electron mobility ratio of 1–2) and exceptional device stability. Devices based on these polymers exhibit outstanding long-term thermal stability at 130 °C for over 200 h and the highest PCE over 9% in a conventional PSC having a single-cell device structure with a ∼300 nm thick active layer.
Polymers | M n [kDa] | PDI | λ onset (film) [nm] | E optg [eV] | HOMOc [eV] | LUMOd [eV] | T d [°C] | T c [°C] | T m [°C] |
---|---|---|---|---|---|---|---|---|---|
a Number-average molecular weight (Mn) determined by GPC with o-dichlorobenzene at 80 °C. b Optical band gap in the film. c HOMO level was estimated from the tangential onset of oxidation (Eonsetox) by cyclic voltammetry. HOMO (eV) = −(Eonsetox − Eonsetferrocene + 4.8). d LUMO level was estimated from the HOMO value and optical band gap of the film. e Decomposition temperature (Td) was determined by TGA (with 5% weight-loss). f Crystallization (Tc) and melting (Tm) temperatures were obtained from the peak maxima by DSC. | |||||||||
PPDTBT | 17.8 | 2.4 | 720 | 1.72 | −5.29 | −3.57 | 396 | 239 | 257 |
PPDTFBT | 29.8 | 2.4 | 720 | 1.72 | −5.35 | −3.63 | 397 | 276 | 283 |
PPDT2FBT | 42.6 | 2.8 | 705 | 1.76 | −5.45 | −3.69 | 402 | 308 | 317 |
Three different types of dialkoxyphenylene and BT-based LBG copolymers were designed by carefully considering the planarity, chain curvature25 and the resulting intermolecular orientations. The noncovalent attractive interactions between S (in thiophene) and O (in alkoxy groups), between S (in thiophene) and F, and between C–H (in thiophene) and N (in BT) minimize the torsional angle, thus maximizing the planarity of the polymer chain (Fig. 1b).13–18,26,27 Noncovalent coulomb interactions have been utilized to increase the planarity and ordering of polymer chains. Guo et al. demonstrated the use of noncovalent S⋯O attractive interactions to fix the chain conformation with improved planarity in methoxy-substituted thiophene and bithiazoles (dihedral angle, ∼0°) compared to unsubstituted (∼22°) and methyl-substituted thiophene-containing analogues (∼68°).17 S⋯F interactions have been emphasized in controlling the stacking orientation in fluorinated benzobisbenzothiophenes by single-crystal X-ray analyses.13 Recently, Ratner and coworkers reported the role of nonbonding interactions in determining conformations of conjugated polymers and small molecules.28 In this paper, the binding energy was calculated to be 2.2, 0.51 and 0.44 kcal mol−1 for CH⋯N, O⋯S and F⋯S nonbonding interactions, respectively. The branched 2-hexyldecyloxy substituents endow great solubility in common organic solvents with little influence on the intermolecular packing in the film by keeping the branching point away from the main chain. Additionally, by changing the number of fluorine substituents, the electronic structures (such as frontier orbital energy levels) of the polymers can be fine-tuned, which significantly influences the thermal, electrical properties and temporal stabilities of the resulting devices.
Computational studies using density functional theory (DFT, Jaguar quantum chemistry software, M06-2X/6-31G** level) were performed.29-33 As shown in Fig. S1†, torsional profiles obtained by the introduction of fluorine atoms (in PPDTFBT and PPDT2FBT) were expected to be similar with that of PPDTBT because of the small size of fluorine atoms and intrachain F⋯S interactions. The introduction of alkoxy substituents on the phenylene ring was observed to decrease the torsional angle (18.4°–20.9°) via the S⋯O noncovalent interaction compared to the alkyl-substituted structure (38.6°). Fig. S1† shows the minimum energy conformations of PPDTBT, PPDTFBT and PPDT2FBT. According to the torsional profiles for PPDT2FBT, there are two minimum energy conformations for the thiophene–dialkoxybenzene linkage. This means that the S⋯O interaction is comparable with that of the O⋯H–C interaction (Fig. S1(b)†). The same argument can be applied to the thiophene–difluoro BT linkage (Fig. S1(a)†). We guess that these minimum energy conformations are expected to repeat randomly in the polymeric backbone for the polymers, as displayed in Fig. 1b. In order to estimate interchain packing interactions, the binding energies were calculated by considering three types of cofacial interactions.34 The head-to-tail (HT) configuration was found to be the most stable among the various possible configurations for all the polymers (Fig. S2†). The calculated binding energies of HT-type cofacial dimers were −18.7, −22.5, and −24.2 kcal mol−1 for PPDTBT, PPDTFBT and PPDT2FBT, respectively. In particular, the introduction of fluorine substituents greatly affected the interchain packing by way of attractive C–F⋯H, F⋯S and C–F⋯πF interactions in the adjacent polymeric chains.35,36 As shown in Fig. S2,† the replacement of methoxy substituents on the phenylene ring with ethyl groups destabilized the structure by ∼3 kcal mol−1 because of the twisting in the main chain caused by the absence of S⋯O or O⋯H–C interactions. The HOMO levels were measured to be −5.29, −5.35 and −5.45 eV for PPDTBT, PPDTFBT and PPDT2FBT by cyclic voltammetry (CV) (Fig. S3†), respectively. The LUMO levels were estimated to be −3.57, −3.63 and −3.69 eV for PPDTBT, PPDTFBT and PPDT2FBT, respectively, from the HOMO values and the optical band gaps of the films. Though the HOMO and LUMO electronic structures were calculated to be similar for the three polymers (Fig. S4†), their energies were clearly stabilized upon fluorine substitution. The resulting energy band structures are also summarized in Fig. 1e.
Fig. 1c shows the normalized UV-vis absorption spectra of the polymers in chloroform and in the film. All the polymers exhibit broad absorption in the range of 350–750 nm with two distinct high and low energy bands attributed to the localized π–π* and internal charge transfer transitions, respectively. In chloroform, the maximum absorption was measured at λabs = ∼575 nm, ∼585 nm and ∼650 nm for PPDTBT, PPDTFBT and PPDT2FBT, respectively. More importantly, the shoulder peak at 650 nm was gradually enhanced with increasing fluorine substitution. In the film, three polymers show similar UV-vis profiles, where the spectra are red-shifted and the shoulder peak is substantially intensified, relative to those in solution. The differences in UV-vis spectra in the solution and film emphasize the facile interchain organization in the solid state. Optical band gaps were determined to be 1.72–1.76 eV for the polymer films. Thermal stability of the polymers was analyzed by thermogravimetric analysis (TGA) and differential scanning calorimetry (DSC) (Fig. 1d). PPDTBT, PPDTFBT and PPDT2FBT showed the decomposition onset temperatures with 5% weight loss at 396, 397 and 402 °C, respectively. Clear melting temperatures (Tm) at 257, 283 and 317 °C, and recrystallization points (Tc) at 239, 276 and 308 °C were measured for PPDTBT, PPDTFBT and PPDT2FBT, respectively, indicating clear crystalline nature of the polymers. This clearly indicates that the introduction of fluorine atoms has a significant effect on the thermal properties of the polymers. Detailed optical, electrochemical and thermal properties of the three polymers are summarized in Table 1.
Polymer | Thermal annealinga | J SC (mA cm−2) | V OC (V) | FF | PCE (%) Best/Ave. |
---|---|---|---|---|---|
a Thermal annealing at 130 °C for 10 min. | |||||
PPDTBT | No | 9.77 | 0.76 | 0.58 | 4.27/4.12 |
Yes | 10.40 | 0.81 | 0.61 | 5.08/4.89 | |
PPDTFBT | No | 10.6 | 0.76 | 0.59 | 4.72/4.38 |
Yes | 10.2 | 0.81 | 0.62 | 5.11/5.02 | |
PPDT2FBT | No | 12.9 | 0.78 | 0.71 | 7.18/6.94 |
Yes | 11.40 | 0.86 | 0.74 | 7.26/7.06 |
Processing additives offer an efficient way to control the morphology of the active layer by selectively solvating one of the components in BHJ systems.38–40 Several additives were tested to modulate the morphology of the blend films prepared from a 1 wt% chlorobenzene (CB) solution, where the clearer additive effects were observed compared to the films from DCB solutions. In contrast to poor performances of the devices fabricated without processing additives (Fig. S6 and Table S3†), devices with additives showed substantial improvements in photovoltaic properties. Among the tested additives, diphenyl ether (DPE) was found to be an appropriate processing additive for our polymers. The addition of DPE to CB (CB:DPE = 98:2 by volume) led to remarkable enhancements in device performances for the three polymeric structures, showing ca. 5–8% PCE values (Fig. 4). PPDTBT:PC70BM and PPDTFBT:PC70BM devices showed the best PCE with a film thickness of ∼170 nm. Interestingly, the PPDT2FBT blend film exhibited the best efficiency using a thick (290 nm) film (as shown in Fig. S7 and Table S4†). Fig. 4a and b show the J–V characteristics and external quantum efficiency (EQE) of the devices with DPE. Table 3 summarizes the detailed photovoltaic parameters. The PPDT2FBT device showed the highest PCE of 8.64% with a JSC of 15.73 mA cm−2, VOC of 0.78 V and FF of 0.71, reaching EQE values of over 80% in the range of 470–550 nm with a maximum EQE of 82.5% at 490 nm (Fig. 4b). These enhancements in JSC and FF by the addition of DPE are closely related to the strong interchain ordering with the recovery of the strong shoulder peak (originating from strong intermolecular packing and/or π–π stacking) of polymer:PC70BM blend films in UV-vis spectra (Fig. 5). In addition, we also measured a remarkable temporal stability of PPDTFBT- and PPDT2FBT-blend films (in CB without DPE) at 130 °C for 200 h, compared to PPDTBT and P3HT based devices (Fig. S8a†). For devices with the processing additive (DPE), poor thermal stability was measured, showing a gradual decrease in PCE, compared to the devices without DPE (Fig. S8b†). It has been recognized previously that thermal treatment induces agglomeration with a concomitant decrease in device performance with processing additives. The solvent additive allows the components to remain partially dissolved, thus affecting the morphology and diffusion rate of fullerene molecules in the polymer matrix, and promoting the growth of fullerene agglomerates. This can accelerate phase separation between the polymer and fullerene moieties during heat treatment, adversely affecting device performance.41
Fig. 4 Photovoltaic characteristics of polymer:PC70BM-based devices fabricated using a CB:DPE solvent mixture. (a) Current density versus voltage (J–V) characteristics and (b) external quantum efficiency (EQE) of polymer:PC70BM-based PSCs. (c) J–V characteristics of the optimized PPDT2FBT:PC70BM-based PSC obtained from our laboratory and (d) certified by the KIER, respectively (c and d: with MeOH treatment). The inset of Fig. 4c shows the EQE values over 80% in the range of 460–570 nm with the maximum EQE of 83.6% at 490 nm. |
Polymer | Active layer thickness (nm) | MeOH treatment | J SC (mA cm−2) | V OC (V) | FF | J SC [cal.]a (mA cm−2) | PCE (%) | |
---|---|---|---|---|---|---|---|---|
Best | Ave. | |||||||
a J SC [cal.], calculated JSC from a EQE curve. | ||||||||
PPDTBT | 170 | No | 11.73 | 0.70 | 0.63 | 11.77 | 5.17 | 5.04 |
PPDTFBT | 175 | 13.29 | 0.73 | 0.69 | 12.88 | 6.64 | 6.45 | |
PPDT2FBT | 290 | 15.73 | 0.78 | 0.71 | 15.59 | 8.64 | 8.39 | |
PPDT2FBT | 290 | Yes | 16.30 | 0.79 | 0.73 | 15.94 | 9.39 | 9.21 |
To further optimize the PPDT2FBT device, the top of the active layer was treated with methanol (MeOH). Solvent treatments can be an effective strategy for simultaneously enhancing all device parameters.42,43Fig. 4c and d show J–V characteristics of optimized, MeOH treated, PPDT2FBT:PC70BM devices as measured in our laboratory and certified by the Korea Institute of Energy Research (KIER), respectively. More than 50 devices were fabricated for device optimization. The best performing device exhibited a PCE of 9.39% (average PCE = 9.21%) with a JSC of 16.30 mA cm−2, VOC of 0.79 V and FF of 0.73 (Table 3). The EQE values of these devices are above 80% in the range of 460–570 nm with a maximum EQE of 83.6% at 490 nm (inset of Fig. 4c). The surface morphologies of PPDT2FBT:PC70BM films with and without MeOH treatments were characterized by AFM (Fig. S9†). There were no observable changes in the AFM images, indicating that the effects of MeOH treatment do not arise from reconstruction of the film surface. Similar data and the detailed studies on the MeOH treatment effects have been reported previously. Bazan and Heeger et al. reported that MeOH treatment enhanced the photovoltaic efficiency by increasing the internal electric field and surface potential by Kelvin probe force microscopy (KPFM) and impedance measurements. Additionally, the series resistance decreased and the shunt resistance increased after methanol treatment, in good agreement with the observed improvements in JSC and FF.42,43 A certified PCE of 8.78% was obtained by the KIER (Fig. 4d and S10†) from a UV-epoxy encapsulated sample. This PCE was ∼5% lower than the average PCE as measured in our laboratory, which could be attributed to non-ideal encapsulation.7,44 The certified results confirm that the measured PCE values of over 9%, as obtained in our laboratory, are reasonable. Furthermore, this work is the first report that showcases efficiency over 9% with a conventional-type, 290 nm thick, single cell structure without any additional interfacial layer (Table S5†). The high PCE and device thickness also suggest a meaningful approach for real commercial applications of PSCs. Although remarkable improvements in PCE have been reported in PSCs, the device thickness is on the order of ∼100 nm. It is of great importance to develop photovoltaic materials which can function effectively at the greater film thickness. It is not currently viable to fabricate uniform and defect-free films on the order of 100 nm thickness using industrial solution casting techniques. Most previous PSCs showed that the performance degrades with the concomitant decrease in FF, with increasing film thickness. This must be closely related to space charge accumulation and charge recombination losses which become stronger with thicker films. It is noteworthy to emphasize that a ∼300 nm thick active layer in the PPDT2FBT:PC70BM device attenuates incident light almost completely without damage in the fill factor (0.71–0.73), showing a high JSC of 15.7–16.3 mA cm−2. These superior properties are closely related to the molecular structures and pronounced crystalline morphology in the film.
Fig. 6 HR-TEM images of polymer:PC70BM films without (a–c) and with DPE (d–f). PPDTBT (a and d), PPDTFBT (b and e) and PPDT2FBT (c and f). |
In order to quantify charge carrier mobilities, hole-only (ITO/PEDOT:PSS/polymer:PC70BM/Au) and electron-only (FTO/polymer:PC70BM/Al, FTO: fluorine-doped tin oxide) diodes were prepared47 using optimized BHJ films (CB:DPE = 98:2 vol%) with various film thicknesses (200–1000 nm) and their J–V characteristics were analyzed by the space charge limited current (SCLC, JSCL) method. The potential loss due to the series resistance of the ITO and the built-in potential were carefully considered in order to ensure accuracy in the measurements. The J–V characteristics show a quadratic dependence on voltage over a range of several volts and an inverse cubic dependence on the film thickness, consistent with the Mott–Gurney relationship (eqn (1)),48,49
JSCL = 9ε0εrμV2/(8L3) | (1) |
The average hole and electron mobilities were determined to be μ (hole) = 3.2 × 10−4, 5.5 × 10−4 and 3.0 × 10−3 cm2 V−1 s−1, and μ (electron) = 2.8 × 10−4, 4.2 × 10−4 and 1.5 × 10−3 cm2 V−1 s−1 for PPDTBT, PPDTFBT and PPDT2FBT devices, respectively. Plots of films with similar thickness are found in Fig. 8, while additional plots with films of various thicknesses are found in Fig. S11 and Table S6.† All the polymers showed well balanced hole/electron mobility ratios in the range of 1–2 with various film thicknesses. PPDT2FBT showed ∼1 order higher hole and electron mobilities relative to other two polymers. Notably, PPDT2FBT showed negligible changes in the carrier mobility even at ∼1 μm film thickness. These SCLC results are consistent with the PCE data of PPDT2FBT, showing no decrease in VOC and FF with a 290 nm thick active layer, resulting in a high JSC of 15.7–16.3 mA cm−2 due to the increased light absorption.
To further understand the detailed film morphology of the three polymers with/without PC70BM and before/after additive treatments, the molecular arrangements and packing characteristics of the thin films were studied by grazing incidence wide angle X-ray scattering (GIWAXS).43,50Fig. 9 shows GIWAXS patterns for pristine polymer and polymer:PC70BM blend films prepared from CB solutions with and without DPE. From the GIWAXS profiles, packing parameters were extracted and are listed in Table S7.† Pronounced reflection peaks of (100), (200), and (300) in the out-of-plane direction were observed in the pristine PPDTBT film, showing a lamellar spacing of 18.9 Å (Fig. 9a and S12†). For the pristine PPDTFBT and PPDT2FBT films, the lamellar spacing calculated from the (100) diffraction peak was slightly increased to 20.7 Å. On increasing the number of fluorine substituents on the BT unit, the in-plane lamellar diffraction peak was also intensified (strongest in PPDT2FBT). Therefore, the presence of fluorine atom may induce a face-on lamellar orientation coexisting with the edge-on lamellar stacks, which may facilitate an effective three-dimensional charge transport. The PPDTBT sample showed no π–π stacking (010) peak in the out-of-plane direction. Interestingly, the (010) peak (d = ∼3.7 Å) in the out-of-plane direction becomes pronounced by the introduction of fluorine substituents. The π–π stacking distance was shorter in PPDT2FBT (3.72 Å) than in PPDTFBT (3.78 Å), which indicates a stronger cofacial interchain orientation between the neighboring chains. Upon addition of DPE to the pristine polymer films, similar trends were observed with increased scattering intensity, indicating more pronounced interchain orientation with DPE. The GIWAXS patterns of polymer:PC70BM blend films are shown in Fig. 9c and d. The lamellar spacing for polymer:PC70BM blend samples was measured to be 19.1–19.9 Å for the three structures, showing similar spacings in the pristine polymer films. Upon addition of DPE, the diffraction patterns are clearer without noticeable changes in the lamellar spacing. We also observed a further reduction in the π–π stacking distance (3.57–3.59 Å) for the PPDTFBT:PC70BM and PPDT2FBT:PC70BM films with DPE, indicating an intensified interchain orientation upon addition of DPE. Interestingly, the blend films with DPE show shorter π–π stacking distances than those of the pristine polymers (PPDTFBT: 3.78 Å and PPDT2FBT: 3.72 Å). The solvent additive allows the components to remain partially dissolved and affects the morphology and diffusive rate of PC70BM in the polymer matrix. This may allow a longer time for polymer chains to self-organize into highly ordered intermolecular structures.51
Footnote |
† Electronic supplementary information (ESI) available: Synthetic details for monomers, DFT calculation, additional GIWAXS and photovoltaic characterization data. See DOI: 10.1039/c4ee01529k |
This journal is © The Royal Society of Chemistry 2014 |