A Hyperbranched Conjugated Polymer as the Cathode Interlayer for High‐Performance Polymer Solar Cells

An alcohol-soluble hyperbranched conjugated polymer HBPFN with a dimethylamino moiety is synthesized and used as a cathode interlayer. A PCE of 7.7% is obtained for PBDTTT-C-T/PC71BM based solar cells. No obvious interfacial dipole is found at the interface between the active layer and HBPFN however, an interfacial dipole with the cathode could be one of the reasons for the enhanced performance. Organic photovoltaics (OPVs) have drawn great attention owing to their key advantages of low-cost, high throughput fabrication, light weight physical characteristics, and the lucrative possibility of integration directly into flexible devices.1 The power conversion efficiency (PCE) of OPVs has rapidly increased in recent years and already reached efficiencies of up to 9.2% with good ambient stability for single cell devices.2 This has been based on high efficient donor and acceptor blend with controlled morphology and improved device structures.2 One key advance for achieving such efficiency values was the use of an alcohol soluble conjugated polymer cathode interlayer. This kind of conjugated polymer cathode interlayer was first used as an electron injection layer in polymer light-emitting diodes3 and then applied to organic field effect transistors4 and recently in OPVs.5-7 In addition, other organic electron collection materials have also been reported, for example, polyfluorenes grafted with K+ intercalated crown ethers (PFCn6:K+),8 fullerene derivatives,9 and insulating polymers with amine group such as polyethylenimine (PEIE).[7],10 Polar solvent treatment can also improve the performance of the organic electronic devices.11 It is believed that the molecular and interfacial dipoles or n-doped conducting nature are the main reasons for the working mechanisms. However, a deeper understanding of how the interface modification works is still needed. Previously, we reported three-dimensional alcohol-soluble conjugated polymers with pendant ammonium salts or neutral amines as cathode interlayers and showed that they work similarly or better than the linear analogues for OPVs.12 However, they have not been applied to high performance OPVs, nor is it clear how they function. Due to the easy synthetic accessibility -typical of one–pot syntheses, hyperbranched polymers have been extensively studied.13 Hyperbranched polymers are highly branched macromolecules with three-dimensional dendritic architecture. They are imperfect architectural relatives of dendrimers but they are very similar to dendrimers for their properties such as excellent solubility, prominent processability, highly branched molecular structure and controllable densities of reactive or non-reactive end-groups. In this communication, we synthesize a novel hyperbranched alcohol soluble polymer HBPFN (Figure 1) by a one-pot "A2+B3" Suzuki polymerization and use it as a cathode interlayer in OPVs with the active blend layer of a high efficiency donor, poly{[4,8-bis-(2-ethyl-hexyl-thiophene-5-yl)-benzo[1,2-b:4,5-b′]dithiophene-2,6-diyl]-alt-[2-(2′-thylhexanoyl)-thieno[3,4-b]thiophen-4,6-diyl]} (PBDTTT-C-T)14 (Figure 1), and [6,6-phenyl C71-butyric acid methyl ester (PC71BM)] acceptor. We demonstrate that incorporating the HBPFN cathode interlayer between the active layer and Al electrode causes a dramatic enhancement in Voc, Jsc and FF values, resulting in an overall PCE enhancement from 4.8% to 7.55% (a 56% relative enhancement). The PCE of 7.55% for the OPV with HBPFN interlayer is ∼10% higher than that of the device with a typical linear interlayer PFN (6.88%). The PCE can be further enhanced to 7.70% by methanol treatment before incorporation of the HBPFN interlayer. The improved hole mobility, reduced dark current, good wetability of the interlayer on the active blend layer and the likely formation of an interfacial dipole with the cathode contribute to the enhanced performance. The chemical and device structures of the studied interlayers and OPVs, respectively, are presented in Figure 1. The synthetic procedure for HBPFN is shown in Scheme S1 in the supporting information (SI) and the detailed synthesis processes are described in SI as well. It was dissolved in methanol in the presence of small amount of acetic acid, but was insoluble in other common organic solvents such as chloroform, THF and toluene. The 1H NMR shows chemical shifts at 7.64–7.78 ppm from protons on the fluorene units, 7.20–7.63 ppm from the triphenylamine unit, 2.85 ppm from the methylene adjacent to the nitrogen atom, 2.55–2.66 ppm from the protons on the sp3 carbon connected to the fluorene unit and 1.08–1.26 ppm from protons at the middle methylene and methyl groups. The ratio of integration is 3:4:2:2:8, which matches with the hyperbranched structure. The molecular weight of HBPFN was not available due to the solubility issue. However, we designed and synthesized a model hyperbranched polymer HBPFO (Scheme S2), in which the side amino- groups on HBPFN are replaced with octyl side chain. GPC analysis showed its weight-average molecular weight (Mw) and polydispersity are 21,900 and 1.44, respectively. Their bandgaps and energy levels including the highest occupied molecular orbital (HOMO) and the lowest unoccupied molecular orbitals (LUMO) were estimated and are listed in the SI. As a comparison, two linear polymers PFN and polydioctylfluorene (PFO) were also studied. The small variations of the HOMO and LUMO levels between the amino-containing materials and the model polymers indicate that the side chain amino- groups do not change the electronic structure significantly. The hyperbranched polymer HBPFN and the linear polymer PFN were applied as cathode interfacial layers in OPVs. The optimized concentrations for spin casting HBPFN and PFN cathode interlayer solutions were found to be at 0.6 mg/mL and 0.4 mg/mL respectively, and the optimized device performance is shown along with other control devices for the current density-voltage (J–V) characteristics of the devices under illumination of AM 1.5G, 1000 W/m2 (Figure 2a) and at dark (Figure 2b), and for the spectral response profiles (Figure 2c). The device structure was ITO substrate/PEDOT: PSS (30 nm)/PBDTTT-C-T: PC71BM (100 nm)/with or without interlayer (5 nm)/Al (100 nm). The extracted device performance metrics, including series resistance (Rs) and shunt resistance (Rsh), are summarized in Table 1. From this table we can see that the PCE increases in the sequence according to the interface modifications: methanol < PFN < HBPFN < methanol/HBPFN. We note that all the interfacial treatments improved the efficiency compared to the control device without interface modification; and a combination of methanol treatment followed by HBPFN interlayer gave the highest PCE. All device parameters are improved significantly from the control device to the best one: PCE from 4.83% to 7.70%, Voc from 0.69 to 0.78 V, Jsc from 14.32 mA/cm2 to 17.15 mA/cm2, and FF from 48.8% to 57.5%. In addition, the series resistance is reduced from 19.86 Ωcm2 to 1.84 Ωcm2 and the shunt resistance is enhanced from 2.86 kΩ cm2 to 10.25 kΩ cm2. The PCE of the device with Ca/Al cathode was 7.30% with Voc of 0.76 V, Jsc of 15.96 mA/cm2 and FF of 60.1%.[9] Collectively, these results show that the performance of OPVs with a methanol/HBPFN interlayer treatment is comparable to that with amine or ammonium iodide-modified fullerene cathode interlayers and the traditional Ca/Al cathode.[9] To understand the effect of the interfacial treatment from a topographic perspective, atomic force microscopy (AFM) was used (see Figure S5 and table S4 in the SI). The original PBDTTT-C-T: PC71BM film exhibits the surface root mean square roughness (RMS) of 2.08 nm. In the presence of HBPFN interlayer the RMS is very similar (RMS = 3.49 nm) both with and without methanol treatment. The RMS is slightly bigger than the pristine film and also larger than that with methanol treatment and PFN interlayer, which are 2.38 and 3.04 nm, respectively. Beside the morphological variation, we also investigated the wetting properties of the films. Contact angle measurements were performed on different surfaces with drops of water, methanol, and interlayer solutions. The images of the drops and the contact angles are shown in Figure S6 and Table S5 in the SI. The contact angles of water on ITO/PEDOT: PSS/PBDTTT-C-T: PC71BM/Interlayer films show that the active layer surface becomes less hydrophobic with the interfacial treatments than the pristine one. The water contact angle for the pristine film is 100o, and it was reduced to 95° after methanol treatment, 70° and 80° with PFN and HBPFN respectively. To see the wetting properties of the interlayers methanol solutions on the active layer, we measured the contact angles of the interlayer solutions on the active layer surface. HBPFN solution gave a contact angle of 10.4°, a little higher than PFN (8.7°) and methanol itself (8.2°). To simulate the condition for the best device, we also spin cast methanol on the active layer first and then dropped the interlayer solutions on the methanol-modified surface. The HBPFN solution gave lower contact angle of 7o while PFN still remained at 8°. The better wetting of the HBPFN on the active layer after methanol treatment may be partially responsible for the improvement in the device performance. In general, the small contact angles of the interlayer methanol solutions on the active layer indicate the good wetting properties and the advantage of the dimethylamino-modified interlayers for use in conventional solar cells. With the interface modifications, the Jsc and external quantum efficiency (EQE) of the devices were changed (Figure 2c). To explore their origins, the charge transport mobilities were approximated using space charge limited current measurements of electron-only and hole-only devices. Their device structures were: ITO substrate/TIPD (12 nm)/PBDTTT-C-T:PC71BM (100 nm)/with or without interlayer (5 nm)/Al (100 nm) and ITO substrate/PEDOT:PSS (38 nm)/PBDTTT-C-T:PC71BM (100 nm)/with or without interlayer (5 nm)/MoOx(3 nm)/Au (40 nm), respectively. Here TIPD represents a titanium chelate which was used for the electron collection layer.15J-V characteristics of single-carrier devices were measured (curves in figures S3 and S4 in the SI), and the mobilities were calculated by fitting to the Mott-Gurney law,16 and are listed in Table 2. From this table, it is clear that the electron mobility remains similar (from 7.2 × 10−4 to 9.2 × 10−4 cm2/Vs) after interfacial modification, but the hole mobility increases by almost one order of magnitude from 8.3 × 10−5 cm2/Vs for pristine film to 7.2 × 10−4 cm2/Vs for the film with HBPFN, the highest among the various surface treatments. The more balanced charge transport in the device is believed to contribute to the enhanced Jsc and FF observed in the solar cells by reducing the build-up of space charges and charge recombination. Reduced surface traps, optical spacer effect, and increased built-in potential might also be the possible reasons. Finally, scanning Kelvin probe microscopy (SKPM) was used to explore if the interfacial dipole exists with various interlayer treatments. SKPM provides the contact potential difference (CPD) between the probe tip and the surface, which for a conductive film is related to a relative difference of the work functions. This technique has also been applied for the study of conjugated polymer films and photovoltaic blend surfaces, although the interpretation is at times not as straightforward.[6],[10],[11] Our SKPM results are summarized in Table 3. Each value was calculated relative to ITO, which was set at a nominal value of 4.70 eV.17 At this reference value, evaporated gold and PEDOT-PSS films possessed work functions of 4.60 ± 0.025 eV18 and 5.25 ± 0.025 eV respectively. The consistency of these values with previous literature encouraged us to study the active layer surfaces treated by methanol or interlayers PFN and HBPFN. Methanol treatment reduced the work function by 0.09 eV, which is similar to literature reports and implies modification of the interface composition and/or defect structure;11 PFN reduced the work function by 0.31 eV, also in good agreement with literature report;[6] however, HBPFN reduced the work function only by 0.05 eV. Considering that the work function shift of PFN has been described to arise from a dipole,[6] it seems surprising that the HBPFN exhibits such a small work function reduction despite also exhibiting a high Voc within solar cells. To further investigate this apparent discrepancy, we measured the work functions of the interlayers on ITO and on evaporated gold. The work functions of PFN on both ITO and evaporated Au were reduced to ∼4.2 eV, indicating interfacial dipole formation. Similarly, the work functions of HBPFN on ITO and on evaporated Au were reduced to 4.47 eV and 4.07 eV, respectively. This indicates that interfacial dipole formed at these "metallic" interfaces do indeed occur. It was reported that insulating amine-polymers can reduce the work functions of metal oxides and metals by the molecular dipole and interfacial dipole.10 Those amine-containing polymers consist of primary, secondary, tertiary amines or hydroxy groups.10 To confirm if tertiary amine only can work similarly, we made an insulating amorphous polymer, PStN, which has only dimethylamino- moiety on the side chain of polystyrene (Scheme S3). Indeed it reduced the work functions of ITO, Au and the active layer significantly, comparable to the PFN with an ordered structure.2 Based on the SKMP results we summarize the energy level landscape of the various devices studied in Figure 3. We propose the importance of two interfaces: one is between the interlayer and the active blend layer and the other between the interlayer and the cathode. While in reality both will be important, the interfacial modification provided by PFN has clear evidence of the former. In contrast, this contribution to the operation of HBPFN layers is negligible. Since both amorphous PStN and ordered PFN form dipoles with the active layer, that no dipole formation is found on hyperbranched polymer HBPFN with the active layer could not be ascribed to its amorphous structure of the HBPFN. For the interlayer-cathode interface, we argue that the known complexation of Lewis bases to metal surface justifies the formation of the interfacial dipole between the interlayer and Al cathode. Notably, it was reported that molecular dipole and interfacial dipole contributed to the work function reduction of electrodes in the same order of magnitude with insulating amine-containing polymer surface modifiers.[10] It would be interesting to know how these two kinds of dipoles contribute to the interfaces at active layer/interlayer and interlayer/cathodes. For completeness, one of the factors not yet considered is the interaction of the evaporated aluminium cathode with the interlayer. Evaporated aluminium on conjugated polymer surfaces has been shown to modify the extent of conjugation at the surface of conjugated polymer films,19 which may influence the interfacial electronic structure and the resulting recombination dynamics. Understanding such effects is outside the scope of the current work; however, based on the above findings we believe that they are likely to explain the observed electrical differences between devices which utilize PFN and HBPFN, and those that only use methanol treatment. In conclusion, an alcohol soluble hyperbranched polymer HBPFN has been synthesized and used as a cathode interlayer for high efficiency OPVs. A PCE of 7.7% was obtained, which was an improvement compared to previously utilized, linearly conjugated, polymeric interlayer PFN and comparable to traditional devices fabricated with a highly reactive Ca/Al cathode. No interfacial dipole was found between the active layer and the interlayer HBPFN but interfacial dipole formation between the interlayer and the cathode could be one of the reasons for the enhanced performance. Device fabrication: The device structure was ITO substrate/PEDOT: PSS/PBDTTT-C-T:PC71BM/with or without interlayer/Al. ITO substrates were cleaned sequentially with detergent aqueous solution, deionized water, acetone and isopropyl alcohol. The cleaned ITO substrates were then treated with UV/ozone at 30 °C for 10 min. PEDOT: PSS thin layer (about 30 nm) was spin-coated on the treated ITO substrate and annealed at 150 °C for 10 min in air. After that, the PBDTTT-C-T: PC71BM (the ratio is 1:1.5 by weight) blend film was spin-coated from its 1,2-dichlobenzene/1,8-diiodoctane (97:3 vol%) mixed solution on top of the PEDOT:PSS thin layer at 900 rpm for 90 s. For the case of methanol solvent treatment, methanol was spin-coated on top of the PBDTTT-C-T: PC71BM active layer at 5000 rpm for 30 s. In the case of PFN and HBPFN interlayers, their solutions with various concentrations in the mixed solvent of methanol/acetic acid (99:1 vol%)) were spin-coated on top of the active layer. Finally, the devices were completed after deposition of 100 nm Al as the electrode in a vaccum thermal evaporator at a pressure of less than 2 × 10−6 Pa. The active area of the OPV device is 4 mm2. For the interlayer thickness, assuming the layer of HBPFN (on glass/PEDOT:PSS/active layer) to be continuous and having a uniform thickness we can estimate a nominal thickness value of 4 – 5 nm using a simple overlayer calculation. The calculation is based on the exponential attenuation of the N 1s photoelectron signal and a value of 3.25 nm for the effective attenuation length of N 1s photoelectrons in organic materials. The measurements of J–V characteristics and EQE: Current density-Voltage (J–V) characteristics of the devices were measured with a computer-controlled Keithley 236 Source Measure Unit. A Xenon lamp coupled with AM 1.5G solar spectrum filters was used as the light source. The illumination intensity of 1000 W/m2 irradiation was calibrated using a standard monocrystal Si reference cell to make sure the strict light intensity. The external quantum efficiency (EQE) spectra were measured by Stanford Research Systems model SR830 DSP lock-in amplifier coupled with WDG3 monochromator and 500W Xenon lamp. Other Characterization methods: 1H and 13C NMR spectra were recorded on a Bruker AV-400X spectrometer operating at 400 and 100 MHz, in deuterated chloroform or deuterated methanol respectively. Molecular weights of the polymers were measured by Gel Permeation Chromatography (GPC) using a Waters 2695 Separations Module, with a Waters 2414 Refractive Index Detector and a Waters 2996 Photodiode Array Detector, a series of four Polymer Laboratories PLGel columns (3 × 5 μm Mixed-C and 1 × 3 μm Mixed-E), and Empower Pro Software. The molecular weights were calibrated with narrow polydispersity polystyrene standards (Polymer Laboratories EasiCal, MW from 264 to 256000), and molecular weights are reported as polystyrene equivalents based on the refractive index detector using THF as the eluent at a rate of 1.0 mL/min at 30 °C. Recycling preparative GPC was performed in chloroform (3.5 mL/min) at room temperature using a JAI LC-9201 separation module, with a RI-50s refractive index detector and a UV-3740 single wavelength detector, and a series of 2 JAI preparative column: JAIGEL-2H (20 × 600 mm) and a JAIGEL-2.5H (20 × 600 mm). The surface potentials and the work functions of the active layer surfaces and other polymer films were measured on Scanning Kelvin Probe Microscopy (Kelvin Probe 5050 Technology) in air. The work functions were achieved from the average value of 200 points for each sample. X-ray photoelectron spectroscopy (XPS) characterizations were performed in an AXIS Ultra-DLD spectrometer (Kratos Analytical Ltd. Manchester) equipped with a monochromated Al Kα X-ray source at a power of 144 W (12 mA, 12 kV). Ionization potentials of the films were measured by Photo Electron Spectroscopy in Air (PESA) on a Riken Keiki AC2 spectrometer with a light power of 5 nW and a power number of 0.5. UV-Vis spectra were measured with Cary 5E UV-Vis-NIR spectrophotometer. The roughness analysis and atomic force microscopy (AFM) images were carried out on a MFD-3D AFM instrument in AC mode with a NSC15/AIBS Si cantilever (resonant frequency around 325 kHz from μ-masch). The contact angles were measured with a CAM 200 (KSV Instrument LID.) and the photos were taken with a BASLER A602f-2 camera. The hole and electron mobilities were calculated by fitting to the Mott-Gurney law, which is expressed by J = 9εrε0μV2/8L3, where εrε0 is the dielectric permittivity of the active layer, L is the thickness of the active layer, and μ is the mobility.16 Under the space charge limited current, we fitted the experimental J–V data with the equation to get the slope of 2 in the logarithmical scales and the mobilities were obtained as field independent values. This work was supported by CSIRO through CAS-CSIRO joint project and the Victorian Organic Solar Cell Consortium (Victorian Department of Primary Industries, Sustainable Energy Research and Development Grant and Victorian Department of Business and the Australian Solar Institute), and Chinese Academy of Sciences, CAS-CSIRO joint project (Project No. GJHZ1124), the Ministry of Science and Technology of China (No. 2010DFB63530). J. Jasieniak acknowledges financial support through the Australian Research Council grant DP110105341. M. L. Lv acknowledges financial support from Chinese Scholarship Council. M. L. Lv and S. S. Li contribute equally to this work. As a service to our authors and readers, this journal provides supporting information supplied by the authors. Such materials are peer reviewed and may be re-organized for online delivery, but are not copy-edited or typeset. Technical support issues arising from supporting information (other than missing files) should be addressed to the authors. Please note: The publisher is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.