High‐Performance Non‐Fullerene Organic Solar Cells Based on a Selenium‐Containing Polymer Donor and a Twisted Perylene Bisimide Acceptor

A novel polymer donor (PBDTS‐Se) is designed to match with a non‐fullerene acceptor (SdiPBI‐S). The corresponding solar cells show a high efficiency of 8.22%, which result from synergetic improvements of light harvesting, charge carrier transport and collection, and morphology. The results indicate that rational design of novel donor materials is important for non‐fullerene organic solar cells.


DOI: 10.1002/advs.201600117
level offsets and to complement light absorption. [ 24,25 ] Thus, short-circuit current ( J sc ) and V oc can be optimized to achieve a better power conversion effi ciency.
Many non-fullerene acceptors have been developed. At the beginning, poly(3-hexylthiophene) (P3HT) was selected as the donor material for the fabrication of non-fullerene solar cells. However, due to its narrow absorption and relatively high-lying occupied molecular orbital (HOMO) energy level, P3HT-based non-fullerene solar cells suffer from low PCEs, mainly because of the low J sc and fi ll factor (FF) values. [ 26 ] Later, donor-acceptor (D-A) alternating copolymers with better absorption were introduced, which showed improved device effi ciencies. [ 27 ] Choosing a suitable donor material to pair with nonfullerence acceptor is critical. Both the energy level alignment and blend morphology need to be synergistically optimized. One successful example was shown by Jenekhe and co-workers, who reported on a 3,4-ethylenedioxythiophene-linked arylene diimide acceptor (DBFI-EDOT). The choice of suitable BDT-based copolymer (PBDTT-FTTE) and thiazolothiazole-dithienosilole copolymer (PSEHTT) that paired with DBFI-EDOT yielded a PCE of 8.5% with a high J sc (15.67 mA cm −2 ) and a high V oc (0.91 V) showing the importance of donor materials selection. [ 17 ] In this contribution, we designed and synthesized a novel polymer donor, where a benzodithiophene (BDT) derivative with large π-conjugated side chain is used as the electron-rich donor subunit and 1,3-di(thiophen-2-yl)-selenopheno[3′,4′: 4,5]benzo [1,2-c]thiophene-4,8-dione is used as the electron-defi cient acceptor subunit (PBDTS-Se). The incorporation of a selenium heteroatom in the conjugated polymers results in a higher photovoltaic performance than its sulfur counterpart, since selenium has a larger and looser electron cloud than sulfur, which improves the intramolecular Se-Se interaction and facilitates carrier transport. [ 28,29 ] This new donor polymer is paired with our recently developed bay-linked sulfurcontaining perylene bisimide (PBI) dimer (SdiPBI-S). Previous studies using SdiPBI-S acceptor yielded a PCE of 7.16%. [ 16 ] The better optimized donor polymer matches well with SdiPBI-S, resulting in improvements in device parameters. A PCE of 8.22%, with a J sc of 12.90 mA cm −2 , V oc of 0.91 V, and FF of 70.0% was found. The improved performance resulted from synergetic improvement of light harvesting, charge carrier transport and collection, and morphology.
The chemical structures of PBDTS-Se and SdiPBI-S are shown in Figure 1 a. The detailed synthesis of PBDTS-Se is shown in Scheme 1 . Electronic-rich dialkylthio-substituted BDT monomer is prepared according to the literature This is an open access article under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in any medium, provided the original work is properly cited.
Non-fullerene acceptors have attracted tremendous interest due to their potential use as alternatives to the ubiquitous fullerene derivatives in bulk heterojunction (BHJ) organic solar cells. [1][2][3][4][5][6] Extensive studies have been carried out to develop polymeric and small molecule acceptors over the past decade. [7][8][9][10][11][12][13][14] A number of high performance systems have been recently reported with power conversion effi ciencies (PCEs) exceeding 8%, comparable to and exceeding BHJ organic solar cells made from fullerene acceptors in performance. [15][16][17][18] The ineffi cacy of tuning frontier energy levels of fullerenes constrains donor material development. [19][20][21][22] Currently, the donor material design must reference the lowest unoccupied molecular orbital (LUMO) level of fullerene derivatives (e.g., [6,6]-phenyl C 71 -butyric acid methyl ester, PC 70 BM) to optimize open-circuit voltage ( V oc ). [ 23 ] Other disadvantages, such as the weak absorption in the visible region, high production cost, and poor photochemical stability, make fullerenes less than an ideal acceptor material for BHJ organic solar cells. In contrast, nonfullerene acceptors are much more versatile in chemistry. The optical properties and the energy levels can be fi ne-tuned by structural modifi cation. And thus there are more opportunities to group donor/acceptor pairs, to form better frontier energy procedures. [ 30,31 ] Dibranched 2-ethylhexyl were linked to selenophene in one step, then a typical acylation reaction was carried out between dialkyl substituted selenophene and 2,5-dibromothiophene-3,4-dicarbonyl dichloride to get a key intermediate (4). PBDTS-Se was prepared through the Stille coupling reaction between the bis(trimethyltin) monomers and Se. PBDTS-Se exhibits good solubility in commonly used solvents, such as chloroform (CHCl 3 ) and o -dichlorobenzene ( o -DCB) at room temperature and thermally stable up to temperatures beyond 287 °C under an inert atmosphere ( Figure S1, Supporting Information).
The UV-vis absorption spectra of PBDTS-Se, SdiPBI-S, and their blend fi lm (1:1, w/w) are shown in Figure 1 c. The two absorption peaks of PBDTS-Se fi lm were located at 585 and 627 nm, which are slightly redshifted from those in solution ( Figure S2, Supporting Information). It should be noted that the absorption of PBDTS-Se is in the wavelength range of 400-700 nm, which complements the absorption spectra of SdiPBI-S, leading to a broad absorption from 300 to 700 nm.  energy offset, 1.42 V, between the HOMO of PBDTS-Se and the LUMO of SdiPBI-S which is preferred for a high V oc . Steadystate photoluminescence (PL) measurements were performed on the neat and BHJ fi lms (1:1, w/w). Compared to PBDTS-Se and SdiPBI-S neat fi lms, the BHJ fi lm showed strong PL quenching up to 98% and 99% comparing to neat fi lm PL, indicative of highly effi cient charge transfer between PBDTS-Se and SdiPBI-S in BHJ fi lms.
In solar cell fabrication, PBDTS-Se:SdiPBI-S BHJ fi lms of different blending ratios and of different 1, 8-diiodooctane (DIO) contents were used. The device parameters are summarized in Table 1 and Table S1 (Supporting Information). The current density-voltage ( J-V ) curves and the corresponding incident photon conversion effi ciency (IPCE) spectra of solar cells are shown in Figure 2 and Figure S4 (Supporting Information), respectively. The optimal PBDTS-Se:SdiPBI-S weight ratio was found to be 1:1. For as cast thin fi lms without using a DIO additive, PBDTS-Se:SdiPBI-S solar cells achieved an PCE of 7.59%, with a V oc of 0.94 V, a J sc of 12.20 mA cm −2 , and a FF of 66.2%. The addition of DIO can further increase the cell effi ciency. When processed from 0.5% DIO, solar cells exhibited the best photovoltaic performance with a V oc of 0.91 V, a J sc of 12.90 mA cm −2 , and a FF of 70.0%, leading to a PCE of 8.22%. We obtained an average effi ciency of 8.01% for more than 20 devices under this condition, indicating the excellent reproducibility of device fabrication. A PCE of 8.22% is among the highest values reported in the literature for non-fullerene organic solar cells, a distinct improvement over our recent report of SdiPBI-S-based solar cells with a PCE of 7.16%, a V oc of 0.90 V, a J sc of 11.98 mA cm −2 , and a FF of 66.1%. By comparison, the improved PCE mainly comes from a higher J sc and a higher FF. The high J sc is ascribed to the high conversion effi ciency of the absorbed photons into electrons in the device. As seen clearly from Figure 2 b, a very high IPCE value of over 60% over a wide optical range from 400 to 660 nm is obtained, showing a maximum peak over 70% at 440 nm. The calculated J sc value (12.62 mA cm −2 ) from IPCE spectrum is in agreement with the value (12.90) measured from the J-V curve.  The high FF of 70.0% indicates a good active layer morphology that enables effi cient charge transport and charge collection in devices. The charge mobilities of PBDTS-Se neat and BHJ fi lms were measured using the space-charge-limited current (SCLC) method. [ 32 ] The hole mobility of PBDTS-Se and the electron mobility of SdiPBI-S in BHJ fi lm are 1.2 × 10 −3 and 3.3 × 10 −3 cm 2 V −1 s −1 , respectively ( Figure S5, Supporting Information). The addition of 0.5% DIO into BHJ fi lms increased both the hole and electron mobilities to values of 1.9 × 10 −3 and 3.5 × 10 −3 cm 2 V −1 s −1 , respectively, which are comparable to the mobility values of their neat fi lms ( Figure S6, Supporting Information). The high mobilities in the BHJ thin fi lms are similar to those of the neat materials, suggesting good network formation for each component. The balanced mobility in BHJ fi lms contributes to the high J sc and FF in solar cells devices.
To study the exciton dissociation and charge collection in solar cells, the photocurrent ( J ph ) versus the effective applied voltage ( V eff ) analysis was measured. J ph can be obtained by subtracting the dark current from the current under illumination and V eff can be obtained by subtracting the applied voltage from the voltage where J ph is 0. [ 33 ] At a high V eff value (≥2 V), J ph is saturated, since the recombination is minimized due to the high internal electric fi eld in the cell. The charge dissociation probability ( P ( E,T )) can be estimated from the value of J ph / J sat . Under the short-circuit and maximal power output conditions, the P ( E,T ) values are 95%, 97%, and 78%, 84% for solar cells with and without 0.5% DIO additive, respectively. The results indicate that both cells (with and without DIO) have a high exciton dissociation rate and a more effi cient charge collection. We noted that J ph showed a linear dependence on light intensity with the slope equal to 1, indicating a very weak bimolecular recombination in the devices. [ 34 ] It is known that the performance of solar cells strongly depends on the morphology of the active layer. We characterized the structure order of PBDTS-Se and SdiPBI-S in neat and BHJ fi lms using grazing incidence X-ray diffraction (GIXD). [ 35,36 ] The 2D diffraction images are shown in Figure 3 a-d, from which the crystalline ordering and crystal orientation in thin fi lms can be accessed. As shown in Figure 3 a (neat PBDTS-Se fi lm), a high intensity diffraction peak is seen in the in-plane direction, at 0.26 Å −1 , corresponding to a distance of 2.41 nm. This peak comes from the (100) packing of the alkyl chain in PBDTS-Se. A strong π-π stacking peak was observed in the outof-plane direction, located at 1.72 Å −1 , giving a π-π stacking of 0.36 nm. Thus, PBDTS-Se assumes a "face-on" orientation. The (100) crystal size is calculated to be 9.98 nm (4.1 stacks) and the π-π stacking crystal size is calculated to be 2.10 nm (5.8 stacks). SdiPBI-S molecule features two PBI aromatic rings tethered by a single bond and, thus, has a twisted molecular geometry, which induces steric hindrance that retards π-π stacking. In the GIXD profi les, a (100) diffraction at 0.35 A −1 is seen, corresponding to a distance of 1.77 nm. This distance correlates well with the hexyl substituent side chains interspacing on PBI moieties. The azimuthal distribution of the (100) peak is uniform, indicating that the SdiPBI-S crystallites are, on average, randomly orientated in thin fi lm. It is observed that PBI-based  crystalline solids usually showed a close π-π stacking distance due to the large aromatic planes. However, in the current case, no obvious diffraction peak is observed in diffraction image. Only a diffusive ring is seen at ≈1.4 A −1 , corresponding to a distance of 4.83 nm. This distance is larger than usual for a π-π stacking and, thus, charge transport relies on charge hopping from one PBI molecule into the adjacent one through headto-center contact. In this light, the transport can be isotropic. When blending PBDTS-Se with SdiPBI-S, crystalline features from both materials are seen. For BHJ fi lm without DIO, the in-plane line cut profi le in low q region showed a broad peak that is a sum of the diffraction from PBDTS-Se and SdiPB1-S. π-π stacking from PBDTS-Se is seen in the out-of-plane direction, and thus the face-on orientation is preserved in the fi lm of the blend. In the in-plane direction, two refl ections located at 0.27 and 0.32 A −1 are seen in the low q region, which arise from the (100) planes of PBDTS-Se and SdiPBI-S. The crystal size for PBDTS-Se, determined from the width of the (100) peak, is 10.44 nm, and for SdiPBI-S it is 4.04 nm. The π-π stacking refl ection at 1.72 A −1 yielded a crystal size of 2.22 nm. The crystal size for SdiPBI-S is signifi cantly smaller in comparison to PBDTS-Se polymer counterparts. It should be noted that the packing distance for these materials are slightly different, which is probably due to the interactions of the donor and acceptor during casting. When DIO additive was used, the crystalline behavior was not signifi cantly changed and similar diffraction profi les are seen. The position of the (100) peak of PBDTS-Se remained constant but the width decreased yielding a larger crystal size of 12.85 nm. The (100) peak of SdiPBI-S peak shifted to 0.31 A −1 with a similar crystal size. However, the π-π stacking PBDTS-Se under DIO processing shifted to 1.74 A −1 , corresponding to a distance of 0.36 nm. Its crystal size also increased to 2.60 nm. Thus, it is evident that DIO improved the interchain packing of PBDTS-Se along both the (100) and (001) (π-π stacking) directions, giving rise to an improved charge transport and, consequently, improved carrier transport and collection effi ciency. The results agree well with the SCLC measurements.
The feature of phase separation of this BHJ blends was studied by using resonant soft X-ray scattering (RSoXS). [ 37 ] The near edge X-ray absorption fi ne structure spectra for PBDTS-Se and SdiPBI-S are shown in Figure S8 (Supporting Information). Similar spectra were seen for PBDTS-Se and SdiPBI-S, since both polymers have similar atomic compositions. Figure S9 (Supporting Information) shows the energy scans around the carbon K-edge and Figure 4 summarizes the scattering profi les at a photon energy of 286.8 eV where the highest contrast was observed. For BHJ fi lm without DIO additive, a slowly decaying scattering profi les is seen, which suggests that a crystalline network structure has not formed. In contrast, when 0.5% DIO is added, the scattering profi le showed a broad maximum from 0.01 to 0.02 A −1 , corresponding to a length scale of 30-60 nm. And thus a network work structure is formed in the BHJ thin fi lm which is also refl ected in the atomic force microscopy (AFM) results. As shown in Figure S7 (Supporting Information), the surface of the PBDTS-Se is dominated by a fi brillar texture. This feature is preserved in the solution cast BHJ blends. When DIO is used, the amplitude of the surface features increased and fi brillar texture is more pronounced.
These results correlate well with the GIXD and RSoXS results, indicating that the DIO enhanced crystallization and phase separation, which led to the high PCE of 8.2%.
In summary, we demonstrated a novel polymer donor named PBDTS-Se. Through the rational molecular design, PBDTS-Se has a broad absorption (from 300 to 700 nm), a high charge carrier mobility of 2.6 × 10 −3 cm 2 V −1 s −1 , and a relatively low HOMO level of 5.34 eV. When combined with the SdiPBI-S acceptor, solar cells have a V oc of 0.91 V, a high J sc of 12.90 mA cm −2 , and a high FF of 70.0%, leading to a PCE of 8.22%, which is among the highest reported values. The broad absorption, high and balanced hole/electron mobility, and good morphologies with small phase-separated domain sizes in the BHJ fi lms indicate that PBDTS-Se donor matches well with SdiPBI-S acceptor, in terms of optical, electronic, and morphological properties. Our results show that even for a give non-fullerene acceptor, rational design of novel donor materials can lead to further improve its photovoltaic performance.

Experimental Section
Solar Cell Fabrication and Characterization : ITO substrates were cleaned subsequently with detergent, deionized water, acetone, and isopropyl alcohol. After 10 min of UV ozone treatment, PEDOT:PSS (Heraeus Clevios P VP A 4083) was spin casted onto the ITO substrate at 4000 rpm for 40 s and then dried at 150 °C for 10 min in air. The thickness of PEDOT:PSS is about 40 nm, determined with a Dektak XT stylus profi lometer. PBDTS-Se was codissolved with SdiPBI-S in 1,2-dichlorobenzene (DCB) with different weight ratio and DIO contents (the concentration of PBDTS-Se is fi xed at 9 mg mL −1 for all mixed solutions). The active layer was formed by spin coating the mixed solution atop PEDOT:PSS layer at 1000 rpm for 90 s. After that, the samples were annealed at 100 °C for 5 min on a hot plate in a glove box. The optimal thickness of the champion cell is about 120 nm, measured on a Dektak XT stylus profi lometer. Finally, a 10 nm thick Ca layer and a 100 nm thick Al were sequentially deposited by thermal evaporation through a shadow mask at a vacuum of 5 × 10 −5 Pa. The active area of the devices was 4.50 mm 2 . During the measurement, an aperture with the area of 3.14 mm 2 was used. Current density-voltage ( J-V ) characteristics were measured in a Keithley 2400 Source Measure Unit. Short-circuit current was measured in an air mass 1.5 global (AM 1.5 G) solar simulator (Class AAA solar simulator, Model 94063A, Oriel) with an irradiation intensity of 100 mW cm −2 , which was measured  Adv. Sci. 2016, 3, 1600117 by a calibrated silicon solar cell and a readout meter (Model 91150V, Newport). IPCE spectra were measured by using a QEX10 Solar Cell IPCE measurement system (PV measurements, Inc.).
SCLC Measurements : Charge carrier mobilities of the neat and BHJ fi lms were measured by using space charge limit current (SCLC) method. The device structures of the hole-and electron-only devices are ITO/ MoO x /BHJ (or neat fi lm)/MoO x /Al and ITO/Al/BHJ (or neat fi lm)/ Al, respectively. The mobility was determined by fi tting the dark current to the model of a single carrier SCLC using the equation: J = 9 ε 0 ε r µV 2 / 8 d 3 , where J is the current density, d is the fi lm thickness of the active layer, µ is the charge carrier mobility, ε r is the relative dielectric constant of the transport medium, and ε 0 is the permittivity of free space. V = V app -V bi , where V app is the applied voltage and V bi is the offset voltage. The carrier mobility can be calculated from the slope of the J 1/2 -V curves.
GIXD Characterization : It was performed at beamline 7.3.3, advanced light source (ALS), Lawrence Berkeley National Lab (LBNL). X-ray energy was 10 keV and the scattering intensity was recorded on a 2D image plate (Pilatus 2M) with a pixel size of 172 m (1475 × 1679 pixels). The samples-to-detector distance was about 300 mm. The incidence angle was chosen to be 0.16°, which was above the critical angle of BHJ thin fi lm but below critical angle of wafer substrate. The samples were prepared on PEDOT:PSS covered Si wafers in a similar manner to the devices. Resonant soft X-ray scattering experiments were carried out in beamline 11.0.1.2, ALS, LBNL. Experiments were done in transmission geometry. BHJ thin fi lms were fl owed in water and transferred onto Silicon Nitride windows fro Norcada Inc. Samples were loaded into a high vacuum chamber (≈10 −7 torr) to avoid carbon contaminations in ambient environment. A series of phonton energy was used in running the experiments and 300 nm polysteren spheres were used as standard to calibrate the beam center and the sample-to-detecter distance.
Instrumentation : AFM images were taken on a Dimension Icon AFM (Bruker) using a tapping mode. UV-vis absorption spectra were measured using a UV-vis spectrophotometer (Shimadzu UV-2700). Photoluminescence spectra were measured using a Shimadzu RF-5301PC spectrofl uorophotometer.

Supporting Information
Supporting Information is available from the Wiley Online Library or from the author.