Efficient Solar Cells Based on an Easily Accessible Diketopyrrolopyrrole Polymer

Abstract

A new easily accessible, high molecular weight, alternating dithieno-diketopyrrolopyrrolophenylene copolymer provides high electron and hole mobilities exceeding 0.02 cm2 V−1 s−1 in FETs and AM1.5 power conversion efficiencies of 4.6% and 5.5% in solar cells when combined with [60]PCBM and [70]PCBM. The performance of the solar cells strongly depends on the use of a processing agent. Semiconducting copolymers with diketopyrrolopyrrole (DPP) units are emerging as interesting materials for optoelectronic applications in field-effect transistors (FETs)1 and organic photovoltaic cells.2-9 Most DPP copolymers used in efficient organic photovoltaic cells contain synthetically elaborate comonomers, such as dibenzosilole, benzodithiophene, dithienosilole, or substituted bithiophenes. Recently we reported on efficient polymer solar cells that use a high molecular weight polymer, PDPP3T (poly[{2,5-bis(2-hexyldecyl)-2,3,5,6-tetrahydro-3,6-dioxopyrrolo[3,4-c]pyrrole-1,4-diyl}-alt-{[2,2’:5’,2’’-terthiophene]-5,5’’-diyl}]), which has electron rich terthiophene segments alternating with electron deficient DPP units along the chain, to reach a small optical band gap.8 Here we present PDPPTPT (poly[{2,5-bis(2-hexyldecyl)-2,3,5,6-tetrahydro-3,6-dioxopyrrolo[3,4-c]pyrrole-1,4-diyl}-alt-{[2,2’-(1,4-phenylene)bisthiophene]-5,5’-diyl}]) (Figure 1) as new member of the DPP-based copolymer family with favorable semiconducting properties that can be synthesized in four simple steps from commercial products. Introduction of the phenyl ring between two thiophene rings adjacent to the DPP unit in the main chain lowers the highest occupied molecular orbital (HOMO) level of the polymer compared to PDPP3T and increases the open circuit voltage. In combination with [6,6]phenyl-C71-butyric acid methyl ester ([70]PCBM) power conversion efficiencies (PCEs) of 5.5% are obtained under simulated AM1.5G (100 mW cm−2) conditions. Synthesis and Structure of PDPPTPT. Reaction conditions: Pd2(dba)3, PPh3, K3PO4, Aliquat 336, H2O, toluene. PDPPTPT was synthesized by a Suzuki cross coupling reaction between a commercially available benzenediboronic ester and a DPP monomer that can be obtained in three steps (Figure 1).1 The use of Pd2(dba)3/PPh3 as a catalyst yielded a polymer that exhibits a limited solubility (<1 mg mL−1) in all organic solvents tested, except in chloroform. The optical band gap of a solid film of PDPPTPT was determined from the onset of absorption at 1.53 eV; see Supporting Information (SI). The optical band gap is virtually identical in chloroform solution, indicating that even at low concentration in the best solvent known to us, PDPPTPT is in an aggregated state. Heating the solution to 60 °C causes minor changes in the spectrum. As a consequence of the tendency to aggregate, accurate molecular weight determination was not possible and gel permeation chromatography (GPC) traces, obtained for different PDPPTPT concentrations in chloroform, were significantly dissimilar (see SI), caused by an increased aggregate content or aggregate size at higher concentration. The energy levels of the frontier orbitals were estimated by cyclic voltammetry, which provided the onsets of oxidation and reduction at 0.25 and −1.57 V vs. Fc/Fc+, respectively (i.e., −5.35 and −3.53 eV vs. vacuum). This indicates a sufficiently large lowest unoccupied molecular orbital (LUMO) level offset (∼0.5 eV) for electron transfer from PDPPTPT to [60]PCBM or [70]PCBM that have an onset of reduction at −1.07 V vs. Fc/Fc+.10 Compared to PDPP3T (with a HOMO at −5.17 eV and LUMO at −3.61 eV vs. vacuum),8 the HOMO level is lower while the LUMO is higher. FETs were fabricated in a bottom-gate bottom-contact geometry. Typical transfer curves are presented in Figure 2a for positive (left) and negative (right) drain biases. PDPPTPT clearly exhibits ambipolar behavior. Charge carrier mobilities of PDPPTPT were extracted in the saturated region and amount to 0.02 ± 0.01 and 0.04 ± 0.01 cm2 V−1 s−1 for electrons and holes, respectively. These characteristics enabled fabricating CMOS-like inverters by combining two identical ambipolar transistors with a common gate (inset of Figure 2b).11 Figure 2b shows the output voltage (VOUT) of such inverter as a function of the input voltage (VIN) at constant supply bias (VDD). From the steepness of the inverter curve a gain of 15 is obtained, which is comparable to state-of-the-art CMOS-like inverters.8, 12 a) Typical ambipolar transfer characteristics of a FET with PDPPTPT as the semiconductor. Channel length and width were 10 and 10 000 μm, respectively. b) Static input-output characteristics of an inverter based on two identical ambipolar FETs. Channel length and width were 10 and 2500 μm, respectively. Photovoltaic cells were fabricated by spin coating a mixed layer of PDPPTPT and PCBM onto a glass/indium tin oxide (ITO)/poly(ethylenedioxythiophene):poly(styrene sulfonate) (PEDOT:PSS) electrode and evaporation of LiF/Al as a back contact, using an optimized 1:2 PDPPTPT:PCBM weight ratio and layer thickness of 80–90 nm. Using pure chloroform for processing the blends, only modest PCEs of ca. 2% were obtained for cells with [60]PCBM and [70]PCBM, due to low photocurrents. Losses in these devices are mainly related to the coarse de-mixing of the polymer:fullerene blend as evidenced by transmission electron microscopy (TEM, Figure 3) and atomic force microscopy. The morphology of the blend can greatly be improved by addition of a cosolvent during processing.2, 13-15 Here 1,8-diiodooctane (DIO) is used, affording a much finer phase separation. a) Power conversion efficiency of PDPPTPT:PCBM cells vs. DIO content in coating solution. b), c), and d) TEM images of PDPPTPT:[60]PCBM layers processed with 0, 25, and 200 mg mL−1 DIO added to CHCl3. e), f), and g) TEM images of PDPPTPT:[70]PCBM layers processed with 0, 25, and 200 mg mL−1 DIO added to CHCl3. The scale bar in the TEM images is 200 nm. The effect of DIO is most clearly visualized by TEM (Figure 3). Mixed films, processed from chloroform without DIO contain large (>200 nm wide) fullerene clusters that show up as dark regions in the bright-field TEM image, owing to the crystallinity the relatively high density of PCBM (∼1.5 g cm−3).16 These coarse morphologies are known to result in poor device performance.2, 17 In films, processed with DIO, large fullerene domains are absent. Instead, elongated structures are observed, which are brighter than the background and therefore attributed to polymer fibers. Despite the fibrillar nature of the polymer in these blends, electron diffraction did not reveal any reflections, besides those of PCBM crystallites, suggesting that the polymer remains essentially amorphous. DIO is a good solvent for fullerenes and is much less volatile than chloroform. As such, it prevents the formation of large PCBM domains during spin coating. Figure 3a shows that addition of 25 mg mL−1 DIO results in optimal performing cells, both with [60]PCBM and [70]PCBM as the electron acceptor. The current density–voltage (J–V) curves of the best cells are shown in Figure 4a. Under simulated AM1.5G conditions, cells fabricated with [60]PCBM exhibit an open circuit voltage (Voc) of 0.79 V, a short circuit current density (Jsc) of 9.3 mA cm−2, and a fill factor (FF) of 0.63, resulting in a PCE of 4.6% (Table 1) (values are averaged for 8 cells, ranging from 4.5%–4.7%). Changing the acceptor to [70]PCBM, which has an increased absorption in the visible part of the spectrum, improved the Jsc to 10.8 mA cm−2. The Voc remains at 0.80 V and also the FF is preserved at 0.65, resulting in a PCE of 5.5% (Table 1) (values are averaged for 8 cells, ranging from 5.5%–5.6%). The external quantum efficiencies (EQE, Figure 4b) display a very sharp onset at the optical band gap of the polymer. The EQE of cells with [70]PCBM is above 40% over a broad spectral range (770–440 nm). Cells made with [60]PCBM exhibit an even higher response (∼45%) in the 600–750 nm region, but due to a lower response below 600 nm, these are outperformed by the [70]PCBM devices. a) J–V curves and b) EQE of optimized PDPPTPT:[60]PCBM and PDPPTPT:[70]PCBM bulk heterojunction solar cells. In a comparison to two related small band gap DPP-based polymers on which we have reported recently,2, 8 PDPPTPT provides an improved efficiency in a solar cell (5.5% vs. 4.0% for PBBTDPP2 (poly[3,6-bis(4’-dodecyl[2,2’-bithiophen]-5-yl)-2,5-bis(2-ethylhexyl)-2,5-dihydro-pyrrolo[3,4-c]pyrrole-1,4-dione])2 and 4.7% for PDPP3T8). The main reason for the enhanced performance of PDPPTPT is an increase in Voc, which is induced by a lowering of the HOMO level combined with somewhat higher optical band gap to prevent reducing the driving force for electron transfer. To investigate the effect of DIO on the photophysics of the polymer:fullerene blends we used photoinduced absorption (PIA). The PIA spectrum of the blend (Figure 5a) shows two distinct peaks at 0.9 and 1.1 eV and a bleaching signal between 1.5 and 2 eV. In the pure polymer (Figure 5b) only the PIA signal at 1.1 eV of the triplet state is detected. The 0.9 eV peak is assigned to polymer radical cations because it corresponds to the absorption band of chemically oxidized PDPPTPT (Figure 5b). Radical cations (polarons) of conjugated polymers generally give two absorption bands,18 but in case of PDPPTPT the low energy band appears to be shifted below the low energy limit of our PIA setup and only the onset of this band can be detected below 0.5 eV. The presence of both triplets and radical cations in the PIA spectrum of PDPPTPT:PCBM can be rationalized in two ways: either not all excitons created in the polymer are quenched, eventually leading to intersystem crossing and triplets, or part of the initially charge separated states recombine to the triplet state.19, 20 Because the intensity of the triplet absorption in the pristine film is significantly lower than in the blend, the triplets in the blend most likely result form charge recombination rather than from direct intersystem crossing from the excitons. Comparing the PIA spectra of mixed films deposited with and without DIO, we learn that DIO affords larger radical cation and triplet signals, whereas their ratio remains essentially constant. This can be caused by a higher initial charge generation efficiency and/or by a longer lifetime for the photoinduced species. a) Photoinduced absorption spectra recorded at T = 80 K (λex = 780 nm) of PDPPTPT:[70]PCBM layers processed from chloroform without (open circles) and with 25 mg mL−1 DIO (closed squares). b) Absorption difference between oxidized and pristine PDPPTPT (closed squares) and PIA of a pristine PDPPTPT (open circles). In conclusion, we have synthesized PDPPTPT in one step from two easily accessible monomers. PDPPTPT has high charge carrier mobilities for electrons and holes and the absorption spectrum has good overlap with the solar spectrum. These characteristics lead to photovoltaic cells with a PCE of 5.5%, when using [70]PCBM as electron acceptor. The performance of the solar cells strongly depends on the morphology of the active layer of the cell, which was optimized by varying the amount of DIO used in the processing. This is in remarkable correspondence with recent results of high efficiency solar cells based on alternating copolymers using benzothiadiazole or thieno[3,4-b]thiophene as electron deficient units in the main chains.13-15 The synthetic procedures for the preparation of PDPPTPT is given in the SI. [60]PCBM (purity >99%) and [70]PCBM (purity >95%) were obtained from Solenne B.V. UV-vis-nearIR spectra were recorded on a Perkin-Elmer Lambda 900 spectrophotometer. Oxidation experiments were performed under an inert atmosphere with thianthrenium hexafluorophophate21 as the oxidant. Cyclic voltammetry was conducted under an inert atmosphere with a scan rate of 0.1 V s−1, using 1 M tetrabutylammonium hexafluorophosphate in ODCB as the electrolyte. The solution was stirred for 72 h at room temperature to ensure enough PDPPTPT was dissolved. The working electrode was a platinum disk and the counter electrode was a silver rod electrode. A silver wire coated with silver chloride (Ag/AgCl) was used as quasi-reference electrode in combination with Fc/Fc+ as an internal standard. The vacuum level of Fc/Fc+ is assumed at –5.1 eV. Tapping mode atomic force microscopy (AFM) was measured on a MFP-3D (Asylum research) using PPP-NCHR probes (Nanosensors). TEM was performed on a Tecnai G2 Sphera TEM (FEI) operated at 200 kV. FETs were fabricated using heavily doped silicon wafers as the common gate electrode with a 200 nm thermally oxidized SiO2 layer as the gate dielectric. Using conventional photolithography, gold source and drain electrodes were defined in a bottom contact device configuration with channel width and length of 10000 μm and 10 μm, respectively. A 10 nm layer of titanium was used acting as an adhesion layer for the gold on SiO2. The SiO2 layer was exposed to the vapor of the primer hexamethyldisilazane for 60 min prior to semiconductor deposition in order to passivate the surface of the dielectric. PDPPTPT films were spun from a chloroform solution at 1500 rpm for 30 s. Freshly prepared devices were annealed in a dynamic vacuum of 10−5 mbar at 100 °C for 72 h to remove traces of solvent. All electrical measurements were performed in a vacuum using an HP 4155C semiconductor parameter analyzer. The reported values are average values over five different devices. Photovoltaic devices were made by spin coating PEDOT:PSS (Clevios P, VP Al4083) onto pre-cleaned, patterned ITO substrates (14 Ω sq−1) (Naranjo Substrates). The photoactive layer was deposited by spin coating a chloroform solution containing 6 mg mL−1 PDPPTPT and 12 mg mL−1 PCBM and the appropriate amount of 1,8-diiodooctane. The counter electrode, consisting of LiF (1 nm) and Al (100 nm), was deposited by vacuum evaporation at ∼3 × 10−7 mbar. The active area of the cells was 0.091 or 0.162 cm2 and no size dependence was found between these two dimensions. For larger areas, starting at 0.36 cm2, resistive losses in the ITO lead to a reduction of FF. J–V characteristics were measured under ∼100 mW cm−2 white light from a tungsten-halogen lamp filtered by a Schott GG385 UV filter and a Hoya LB120 daylight filter, using a Keithley 2400 source meter. Short-circuit currents under AM1.5G conditions were estimated from the spectral response and convolution with the solar spectrum. The spectral response was measured under simulated 1 sun operation conditions using bias light from a 532 nm solid state laser (Edmund Optics). Monochromatic light from a 50 W tungsten halogen lamp (Philips focusline) in combination with monochromator (Oriel, Cornerstone 130) was modulated with a mechanical chopper. The response was recorded as the voltage over a 50 Ω resistance, using a lock-in amplifier (Stanford research Systems SR830). A calibrated Si cell was used as reference. The device was kept behind a quartz window in a nitrogen filled container. This method has shown to give PCEs that are in close correspondence with those obtained using a WXS-300S-50 solar simulator and using spectral mismatch correction.2 In the J–V measurements made with the white light set up described above, the measured Jsc differed on average by 3%–4% from the value obtained by convoluting the spectral response with AM1.5G (see Table 1). The thickness of the active layers in the photovoltaic devices was measured on a Veeco Dektak 150 profilometer. PIA samples were prepared by spin coating the appropriate mixture on quartz substrates. PIA spectra were recorded by exciting with a mechanically modulated B&W, Tek Inc. laser (λ = 780 nm) pump beam and monitoring the resulting change in transmission of a tungsten-halogen probe light through the sample (ΔT) with a phase-sensitive lock-in amplifier after dispersion by a grating monochromator and detection, using Si, InGaAs, and cooled InSb detectors. The pump power incident on the sample was typically 21 mW with a beam diameter of 1 mm. The PIA (ΔT/T) was corrected for the photoluminescence, which was recorded in a separate experiment. Photoinduced absorption spectra and photoluminescence spectra were recorded with the pump beam in a direction almost parallel to the direction of the probe beam. Temperature of the substrates was controlled by using an Oxford Optistat continuous flow cryostat. Supporting Information is available online from Wiley InterScience or from the author. We would like to acknowledge Tom Geuns (MiPlaza Eindhoven) for preparing the OFET structures. The research was supported by a TOP grant of NWO-CW and is part of the Joint Solar Programme, co-financed by FOM, CW, and the Foundation Shell Research. The work of D.D.N. and V.S.G. is part of the research program of the Dutch Polymers Institute (DPI projects 631 and 660). S.G.J.M. acknowledges financial support from the Dutch Technology Foundation STW. Detailed facts of importance to specialist readers are published as ”Supporting Information”. Such documents are peer-reviewed, but not copy-edited or typeset. They are made available as submitted by 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.