Precise control over reduction potential of fulleropyrrolidines for organic photovoltaic materials †

Organic photovoltaic cells based on two types of organic materials (acceptor and donor) have attracted considerable attention for their low-cost fabrication, and potential for realization of ﬂ exible and light weight devices. For many years organic photovoltaic cells have relied on [6,6]-phenyl C 61 butyric acid methyl ester, a fullerene derivative that is used as an electron acceptor material. A few reports on bisadduct and C 70 derivatives have shown some improvements in device performance; however, further enhancements based on improvements to the fullerene acceptor component have proven challenging. Here we described the device performance of improved acceptor fullerene materials that allow the open circuit voltage to be ﬁ ne-tuned in organic photovoltaic cells to provide high power conversion e ﬃ ciency. Our new approach to designing fullerene materials will accelerate development of n-type semiconductor materials and allow for new low cost organic photovoltaic cells.


Introduction
Environmentally friendly energy generation is an important global issue for society.Solar energy harvesting has been recognized as an effective way of addressing our increasing energy needs.][3][4][5] Current state-of-the art OPV cells are fabricated as a thin lm comprising a mixture of p-and n-type organic semiconducting materials. 1,2This organic semiconducting lm features a nanoscale phase-segregated structure known as a bulk-hetero junction (BHJ), which provides an effective pathway for a charge generation and separation in the lm.
The p-type semiconductors used in such devices are pconjugated aromatic compounds based on polymers and small molecules, which absorb light leading to generation of excitons in the lm. 1,2The absorption of the p-type materials is related to their chemical structures and directly affects their performance in devices.[8][9][10] In contrast to the remarkable development of p-type donor materials, for n-type acceptor materials there have been few advances over [6,6]-phenyl C 61 butyric acid methyl ester (PC 61 BM), which has dominated reports in this eld over the last decade. 11In the term, indene bis-adduct C 60 derivative (IC 60 BA), which is one of the important fullerene derivative, presented an effect to changing a LUMO energy level contributing to obtain a high voltage in OPV cells. 125][26][27][28][29][30][31] Although interest has shied to non-fullerene derivatives, it remains important to develop new fullerenes as low-cost acceptor materials that may allow for commercially viable OPV modules.
We have previously reported the synthesis and application of new fulleropyrrolidine derivatives having various substituents. 32ur previous study indicated that N-phenyl fulleropyrrolidine and its OPV cells showed a higher reduction potential and better device performance, respectively, than those based on PC 61 BM.
In this paper, we show that the reduction potential of fulleropyrrolidine derivatives can be controlled based on the structure of the substituent groups.These modications can in turn control and enhance the open circuit voltage (V oc ) of OPV cells.Our newly synthesized fulleropyrrolidine derivatives were designed based on theoretical calculations.The synthesized materials were characterized by UV-vis absorption, cyclic voltammetry and OPV cell measurements.Relationships between the substitution pattern and V oc value were analysed in detail.An OPV cell based our C 60 -fulleropyrrolidine derivative showed high power conversion efficiency (PCE) of 7.30% with a V oc of 0.80 V.This fulleropyrrolidine represents a next generation fullerene acceptor, which may advance OPV performance beyond that which is possible with PC 61 BM.

Results and discussion
Our new C 60 -fulleropyrrolidine derivatives are illustrated in Fig. 1.Density functional theory (DFT) calculations of the fulleropyrrolidines were conducted, with Gaussian 09 at the B3LYP/ 6-31G level, to guide the design of novel fulleropyrrolidine derivatives before their synthesis as shown in Fig. 2 and 3. To simplify the calculations and ensure no-conjugation between the substituent group and the fullerene p-system, the calculations were performed about the substituent group and its surrounding structure.
The results of the calculated highest occupied molecular orbitals (HOMO) and lowest unoccupied molecular orbitals (LUMO) were related to the structures of the derivatives.When a uorine (F) atom was introduced at the ortho position of the Nphenyl group (compounds 1 and 2), the HOMO energy level was unchanged; however, compound 2 has a higher LUMO level than that of 1.When an F atom was introduced at the ortho position of the C2 phenyl group (compounds 3 and 4), the energy level changed in a similar way to that for compounds 1 and 2; however, the effects were more pronounced in compounds 3 and 4, which had lower LUMO energy levels.These calculations indicate that the energy levels were mainly affected by the number and position of the F atoms that were introduced into the N-and C2-phenyl groups.
When a methoxy group was introduced at the C2-phenyl group instead of an F atom, the energy levels of 5, 6 and 7 shied higher, as observed for the introduction of an F atom into the N-phenyl group (Fig. 3).In compound 7, the combined effects of both the F atoms and the methoxy groups gave the highest calculated LUMO energy level in this series.Based on the LUMO level of our unsubstituted compound (PNP), 32 we could estimate a V oc value of 0.82 V for 7 and 0.80 V for the mono-methoxy compound 6.Notably these small changes in substitution might have a considerable inuence on the LUMO level of the fulleropyrrolidine derivative but did not result in any break down of the fullerene p-system.Maintaining the fullerene p-system is important for forming an ideal BHJ structured thin lm and for obtaining material that is stable under electron injection conditions.
We synthesized compounds shown in Fig. 1, 1-7 from corresponding amino acids and aldehyde derivatives under one   step Prato reaction conditions. 33,34The full synthetic procedures are described in ESI (Fig. S1 †).
UV-vis absorption spectra for fullerene derivatives, 1-7, were measured in chloroform solution (Fig. 4).An absorption band between 250 and 400 nm with a maximum located at 257 nm was observed in all compounds.A sharp peak at 420 nm, which is also observed for PC 61 BM, was assigned to the 58p-system of the fullerene derivative.A weak absorption at 710 nm was attributed to p-p* transitions of the fullerenes.Small absorption differences were observed in the region between 430 and 600 nm, indicating that the substituent groups on the fullerenes had little effect on the electronic structure of the fullerene psystem.
Cyclic voltammetry (CV) measurements were performed in a three-electrode cell to estimate the LUMO energy level of the derivatives from their reduction potential.The results obtained are summarized in Table 1.For comparison, the CV value of PNP 32 is also listed.Reversible redox processes were observed in all the fulleropyrrolidine derivatives, as shown in Fig. S2 †.The following equation was used to calculate the LUMO levels: , where E 1 1/2 is the rst half-wave reduction potential relative to ferrocene (Fc). 35he reduction potentials of the fulleropyrrolidines varied depending on their substitution pattern as shown in Table 1.The LUMO energy level was raised by around 0.02 eV per F atom from 1 to 2. The F atoms on the C2-phenyl group had little effect on the LUMO level and raised the energy by less than 0.01 eV per F atom.The trends of the CV results were consistent with the trends of our DFT calculations, as shown in Fig. 2. The DFT calculations indicated that the methoxy-substituted compound 7 would have the highest LUMO energy level, and indeed, 7 featured the most negative reduction potential in the series, as determined by CV.Notably there were no p-orbital interactions between the substituent group and the fullerene p-system.The differences in the reduction potentials were likely caused by through space p-p orbital interactions between the fullerene and the C2-phenyl ring.In a recent report, Swager and coworkers described a relationship between the cyclobutadiene structure and the fullerene p-system, which resulted in a higher LUMO level. 36ur CV results, suggest that the reduction potential can be controlled by introducing substituents that have through space interactions, but do not disrupt the fullerene p-system.
OPV cells were fabricated from the newly synthesized fulleropyrrolidine derivatives as acceptor materials and PTB7 as a donor material to evaluate the potential of the materials.A device structure of ITO/PFN/PTB7:fullerene derivative/MoO x /Al under simulated AM 1.5G solar irradiation at 100 mW cm À2 was used.The active layer was spin-cast from chlorobenzene (CB) solutions for compounds 1, 2, 5 and 6, and o-dichlorobenzene (DCB) solutions for 3, 4 and 7 owing to the compounds different Fig. 4 UV-vis absorption spectra of fulleropyrrolidine derivatives.solubilities.The current density-voltage (J-V) characteristics and the external quantum efficiency (EQE) spectra are shown in Fig. 5.The OPV parameters are summarized in Table 2.As reported by Janssen and co-workers PC 71 BM-based OPV devices generally show higher performance than those based on PC 61 BM, which has been attributed to the broader absorption of PC 71 BM than that of PC 61 BM. 37 However, C 70 fullerene derivatives are expensive, which makes them less well suited to commercial production of OPV modules.We believe that development of C 60 fullerene-based acceptors is important to realizing economically viable OPV cells.The data reported in Table 2 are comparable to reports on the performance of OPV cells with PC 71 BM as an acceptor. 38,39he J-V curves show that the devices generated high short circuit current density (J sc ) values.High J sc values were achieved in OPV cells based on compounds 2 and 5.The device featuring compound 7 had a slightly lower J sc , which was attributed to a thinner active layer ($90 nm) in this device and reduced photo absorption.The OPV cell with 7 also had a slightly higher series resistance than the others and a relatively low ll factor.
We focused on the V oc of the OPV cells based on our new acceptors.Our CV measurements and calculations revealed a trend in the reduction potential of the compounds.The data in Table 2 and Fig. 5b clearly show that the V oc values reected the different reduction potentials of the derivatives.Relationships between the chemical structure and V oc are illustrated in Fig. 6.
In changing the basic PNP structure (0.742 V, À3.672 eV), by introducing uorine atoms in compounds 1 and 2, we observed that the V oc increased by 0.02 eV per F atom and enhanced the V oc to 0.78 V.For the case of C2-phenyl substitution the measured V oc values were slightly lower than those of 1 and 2, even in compound 4, which had two F atoms.The C2-phenyl uorination had little positive effect on the V oc value.A tetra-uoro compound (4F in ESI †) was also synthesized.The 4F showed above trade off relations of introducing F atoms in phenyl rings, resulting in not so high V oc of 0.780 V.
For the methoxy-substituted series, V oc increased to ca. 0.77, 0.80 and 0.81 for the OPV cells based on 5, 6 and 7, respectively.This trend also agreed with the DFT calculations and CV results.Thus, the V oc in cells with PTB7 (typically 0.74 V) was increased to more than 0.80 V without any disruption to the pristine fullerene p-system.We showed that modifying the substitution pattern is a promising method to ne tune the LUMO level of fullerene derivatives and achieve the best alignment of energy levels in a cell.

Conclusions
In conclusion, we designed new fulleropyrrolidine acceptor molecules guided by theoretical calculations.On the basis of these calculations we synthesized new fulleropyrrolidine derivatives with controlled reduction potentials.The trend in the LUMO levels of the compounds was determined by DFT calculations and was consistent with the results of CV measurements.We fabricated OPV devices and evaluated the performance of the compounds.Introduction of an F atom into the N-phenyl group contributed to enhanced V oc of OPV cells.
Conversely, an F atom on the C2 phenyl group made little contribution to the V oc of OPV cells.The position and number of the F atoms introduced into the phenyl group affected the shiing of the LUMO level.Furthermore, compounds having methoxy-substituted C2 phenyl groups were used as acceptors in the active layer of OPV cells.The energy level of these derivatives also affected the V oc value of the OPV cells.A V oc as high as 0.81 V was achieved with PTB7 as the donor material.In this fulleropyrrolidine series, we were able to tune the V oc value by 0.01-0.07V without breaking the C 60 fullerene p-system.We achieved a high PCE of 7.30% using our C 60 fulleropyrrolidine derivatives.To date, development of new p-type materials with improved light absorption, have mainly contributed to improved energy conversion efficiency in OPV.In this report, we successfully improved the PCE of OPV cells with fulleropyrrolidine acceptor materials, by enhancing the V oc of the cell.This is an important step that may contribute to new gains in already highly optimized OPV cells.

Experimental
Materials and methods Materials.Reagents were purchased from Wako Pure Chemical Industries, Tokyo Kasei Chemical Industries, Merck, and Aldrich, and used without further purication.C 60 fullerene was purchased from Honjo Chemical Corporation.PTB7 as a donor material for the OPV cells was purchased from 1-Material Chemscitech and was used as received.PFN 40 was synthesized in accordance with literature procedures.N-(2-Fluorophenyl)glycine and N-(2,6-diuorophenyl)glycine ethyl ester were prepared according to reported methods. 41,42eneral measurement and characterization.UV-visible spectra were recorded on a Shimadzu UV-3100PC.All spectra were obtained in spectroscopic grade solvents with quartz cells with a 1 cm path length. 1 H and 13 C NMR spectra were recorded with a Bruker Avance III (700 MHz) spectrometer in chloroformd [chemical shis in parts per million (ppm) downeld from tetramethylsilane as an internal standard for 1 H and 13 C]. 19F NMR was recorded with a JEOL JNM-ECS400 (400 MHz) spectrometer in chloroform-d [chemical shis in parts per million (ppm) up eld from uorotrichloromethane as an internal standard for 19 F].Cyclic voltammetry was performed on a BAS CV-50W voltammetric analyzer with Pt working and counter electrodes in chlorobenzene/CH 3 CN (5/1, v/v) solution containing 0.1 mol L À1 Bu 4 NPF 6 .Column chromatography was performed with silica gel, (Kanto Chemical, silica-gel 60N, 40-50 mm).The fulleropyrrolidine compounds were further puri-ed by recycling gel-permeation liquid chromatography (GPC) with a Japan Analytical Industry LC-908 equipped with JAI-GEL 1H/2H columns (eluent: CHCl 3 ) and with Cosmsil Bucky-prep® Columns (eluent: toluene).Elemental analyses were performed on a Perkin Elmer LS-50B by the Elemental Analysis Section of Comprehensive Analysis Center (CAC), ISIR, Osaka University.Surface morphologies of the deposited organic lms were observed with an atomic force microscope (Shimadzu, SPM9600).The thicknesses of deposited organic lms were measured with a KLA-Tencor alpha-step IQ surface measurement proler.
OPV device fabrication and measurements.All cells were fabricated on 150 nm thick ITO-coated glass substrates that were detergent and solvent cleaned.For the inverted device structure, ITO/PFN/polymer : acceptor (1 : 1.5)/MoO x /Al, the PFN interlayer material was dissolved in methanol (2 mg mL À1 ) and spin-coated on top of the clean ITO substrate based on previous reports. 9The organic active layer, with a thickness of 90-100 nm, was prepared by spin-coating a blended solution (25 mg mL À1 ) in chlorobenzene (for 1, 2, 5 and 6) with 1,8diiodoctane (3%, v/v) or dichlorobenzene (for 3, 4 and 7) with 1,8-diiodoctane (3%, v/v) at 1000 rpm for 120 s in a nitrogen atmosphere.A 10 nm MoO x layer and an 80 nm Al layer were then evaporated through a shadow mask to dene the active area (0.09 cm 2 ) of the devices and form the top electrode using EO-5 metal evaporation chamber (Eiko engineering co.Ltd.).Current density-voltage characteristics of the photovoltaic cells were measured in the dark and under simulated solar light, with a Keithley 2400 source meter and a XES-301S solar simulator (San-Ei Electric Co., Ltd.), calibrated to produce 100 mW cm À2 AM 1.5G illumination.All device measurements were performed in N 2 lled measurement-apparatus at room temperature.

Fig. 6
Fig. 6 Relationships between chemical structure and obtained V oc .The values in brackets are the LUMO levels calculated from CV measurements.