Research Article www.acsami.org
High-Efficiency Nonfullerene Polymer Solar Cell Enabling by Integration of Film-Morphology Optimization, Donor Selection, and Interfacial Engineering Xin Zhang, Weiping Li, Jiannian Yao, and Chuanlang Zhan* Beijing National Laboratory of Molecular Science, CAS Key Laboratory of Photochemistry, Institute of Chemistry, Chinese Academy of Sciences, Beijing 100190, P. R. China S Supporting Information *
ABSTRACT: Carrier mobility is a vital factor determining the electrical performance of organic solar cells. In this paper we report that a high-efficiency nonfullerene organic solar cell (NF-OSC) with a power conversion efficiency of 6.94 ± 0.27% was obtained by optimizing the hole and electron transportations via following judicious selection of polymer donor and engineering of film-morphology and cathode interlayers: (1) a combination of solvent annealing and solvent vapor annealing optimizes the film morphology and hence both hole and electron mobilities, leading to a trade-off of fill factor and short-circuit current density (Jsc); (2) the judicious selection of polymer donor affords a higher hole and electron mobility, giving a higher Jsc; and (3) engineering the cathode interlayer affords a higher electron mobility, which leads to a significant increase in electrical current generation and ultimately the power conversion efficiency (PCE). KEYWORDS: perylene diimides, film-morphology optimization, interfacial engineering, solution-processed, nonfullerene organic solar cell
■
INTRODUCTION Nonfullerene organic solar cells (NF-OSCs) utilize n-type organic molecules instead of traditional fullerenes1,2 as the blend acceptor(s) and can be fabricated via easy-handling solution-processed methods. Because the organic acceptor is, with respect to the traditional fullerene one, easily accessible via organic synthesis and easily tunable on its optoelectronic property and aggregation behavior, much effort has been recently made on the explorations of organic acceptors. So far, several kinds of efficient organic small-molecule acceptors such as twisted perylene diimides (PDIs),3−6 twisted tetraazabenzodifluoranthene diimides (BFIs), 7 and twisted bis(diphenylethynyl-tetraphenylazadipyrromethene) zinc(II) complex,8 among others, have been reported.9 Otherwise, several polymer acceptors have been reported, for example, those with PDI10 or naphthalene diimide (NDI)11 as the conjugated electron-accepting moieties. With these small molecules or polymers as the nonfullerene acceptor, the solar-to-electrical power conversion efficiencies of 6−7% were reported by several groups,6,11−19 and very recently a power conversion efficiency (PCE) of 7.1%20 and 7.7%21 was reported from a solutionprocessed twisted sulfur-fused dimeric PDI and a NDI− selenophene conjugated polymer acceptor, respectively. All these exciting results point out the potentiality of NF-OSCs. In an NF-OSC device, both an organic acceptor and organic donor are blended together as the photoactive layer to capture and convert the solar photons into mobile charges, which are then collected by the right electrodes. Absorptivity, filmmorphology, and organic-to-electrode interlayers play very important roles in determining a solar cell’s electrical © 2016 American Chemical Society
performance, among which the latter two are deeply related to the carrier transportation. Organic acceptors such as PDIs normally show absorption in the visible region. The first consideration is to complement the visible absorption spectrum of the acceptor by selecting a low band gap donor material, and the absorption of the donor material will have an impact on the solar photons capturing the ability of the blended film, affecting the solar cell performance. For example, it was observed that the solar cell electrical performance can be significantly improved as the blend donor was changed from P3HT to a low band gap polymer such as PBDTTT-C-T/PffBT4T-2DT with the PCE going from 1.3%/2.4% to 4.0%/5%.14,22,23 Again, a similar phenomenon with the PCEs varying between 1% and 4.5% was observed from several polymer donors with absorption bands covering different spectral wavelength regions.24 Charge dissociation is the second consideration in the selection of donor-to-acceptor combination. It is accepted that the donor-to-acceptor energy offset between the lowest unoccupied molecular orbitals (LUMOs) and that between the highest occupied molecular orbitals (HOMOs) should be larger than 0.3 eV, acting as the driving forces for efficient charge dissociation through the electron and hole transfer paths. Nevertheless, there were only several all-organic combinations as yet realized to show high performance. The third consideration is film morphology. Optimal nanoscale and interpenetrating networks of donor-to-acceptor Received: April 5, 2016 Accepted: June 1, 2016 Published: June 1, 2016 15415
DOI: 10.1021/acsami.6b03926 ACS Appl. Mater. Interfaces 2016, 8, 15415−15421
Research Article
ACS Applied Materials & Interfaces
Figure 1. (a) Molecular structures of the selected polymer donors and the nonfullerene acceptor of bis-PDI-T-EG. (b) Normalized UV−vis absorption spectra of the pure donor and acceptor film and solar photon flux at AM1.5 conditions (black). (c) HOMO and LUMO energy levels which were obtained from ultraviolet photoelectron spectroscopy (UPS) data (Figure S1) and film absorption spectrum (b), respectively.
Table 1. Electrical Parameters of the Solar Cells Obtained with PTB7-Th as the Donor and under Conditions A, B, C, and B+C, Respectively polymer
condition
time [min]
PTB7-Th
A B B+C C
120 45 + 0 30 + 15 0 + 45
Voc [V]a 0.884 0.870 0.876 0.877
± ± ± ±
0.008 0.008 0.009 0.006
Jsc [mA cm−2]a 9.17 11.29 10.89 9.12
± ± ± ±
0.21 0.13 0.22 0.23
FFa 0.430 0.458 0.523 0.549
± ± ± ±
0.016 0.017 0.013 0.011
PCE [%]a 3.47 4.46 4.96 4.36
± ± ± ±
0.24 0.16 0.24 0.27
(3.72) (4.64) (5.23) (4.66)
μeb/μhc [cm2 V−1 s−1] / 8.63/17.2 × 10−5 9.01/12.4 × 10−5 4.73/6.89 × 10−5
a Average values from 10 devices with the best PCE shown in brackets. bElectron mobilities were measured with devices of ITO/TIPD/active layer/ Ca/Al. cHole mobilities were measured with devices of ITO/PEDOT:PSS/active layer/Au.
(Figure 1a)4 as the nonfullerene small-molecule acceptor. Optimizations of film morphology of the PTB7-Th blend with a combination of SA and SVA treatments optimized the hole and electron mobilities and led to a trade-off between Jsc and FF, affording a PCE of 4.96 ± 0.24%. The PCE value was improved to 5.65 ± 0.20% as the cathode interlayer was changed from Ca to PDINO34,35 (Figure 1a). As PBDT-TS1 was selected as the polymer donor, a PCE of 6.22 ± 0.16% and 6.94 ± 0.27% was achieved with Ca and PDINO as the cathode layer.
phase-separated domains are beneficial for charge dissociation and carrier transportation. For an NF-OSC, the morphology of acceptor phase domains is likely an important factor that limits the electrical performance of NF-OSCs. Film-processing techniques which were proven efficient for fullerene OSCs have been widely shown to work for optimizations of film morphology of all-organic systems. For example, it was shown that processing the blend film with high boiling point solvents as additives can optimize the film morphology and/or with solvent or solvent-vapor annealing (SA or SVA) techniques can reconstruct the film morphology, both of which lead to formation of charge-dissociation favorable and charge-transport beneficial film morphologies with improved electron transport.13,20 Again, it was indicated that (1) well-ordered PDI domains suffer from poor electrical coupling between the PDI domains, which leads to a decreasing electron mobility,25 and (2) phase size of acceptor domains in the phase-separated blend films could be correlated to the short-circuit current density (Jsc) in some PDI:small-molecule donor systems.26 Besides, the fourth consideration is to modify the organic-toelectrode interfaces, the cathode interlayers, which have been proven effective in improving the collection of electrons, and so on raising the Jsc, fill factor (FF), and ultimately the PCE value of fullerene OSCs.27−30 In this paper, we selected three low band gap polymers, PTB7,31 PTB7-Th,32 and PBDT-TS133 (Figure 1a), as the donors and a previously reported PDI dimer of bis-PDI-T-EG
■
RESULTS AND DISSCUSSION Figure 1b gives the absorption spectra of the three polymers and the PDI acceptor. The absorption band red-shifted to a longer wavelength region for the polymer, going from PTB7 to PTB7-Th and PBDT-TS1, which became more complementary to the absorption of the PDI acceptor, and the all-organic combination had a wider coverage of solar emission spectrum. All three polymers had a higher-lying LUMO or HOMO energy level than the PDI acceptor (Figure 1c and Figure S1), providing driving forces for efficient dissociation of the excitons generated by the blended donor and acceptor domains. First, we selected PTB7-Th as the blend donor. High boilingpoint additive solvent of 1,8-diiodooctane (DIO) and post treatment following a SA/SVA process were used for filmmorphology optimizations. As the as-cast photoactive layer was put in an open Petri dish for 2 h (without any treatments of SA 15416
DOI: 10.1021/acsami.6b03926 ACS Appl. Mater. Interfaces 2016, 8, 15415−15421
Research Article
ACS Applied Materials & Interfaces
Figure 2. Experimental J−V characteristics (a) and EQE responses (b) of the optimal solar cells with PTB7-Th as the donor. The out-of-plane and in-plane 1D-GIXRD data (c) of the pure PTB7-Th film (green) and the solar cell blends obtained under conditions B, B+C, and C, respectively.
Table 2. Electrical Parameters of the Solar Cells Obtained with PTB7, PTB7-Th, and PBDT-TS1 as the Donor Materials, Respectively polymer
cathode
PTB7 PTB7-Th PBDT-TS1 PBDT-TS1 PBDT-TS1 PBDT-TS1 PBDT-TS1
PDINO/Al PDINO/Al PDINO/Al Ca/Al Al MeOH/Ca/Al MeOH/Al
Voc [V]a 0.840 0.876 0.887 0.885 0.874 0.885 0.865
± ± ± ± ± ± ±
0.006 0.007 0.005 0.008 0.007 0.006 0.005
Jsc [mA cm−2]a 8.46 11.14 13.23 12.41 11.66 11.85 11.25
± ± ± ± ± ± ±
0.19 0.21 0.19 0.22 0.23 0.18 0.26
FFa 0.552 0.583 0.594 0.569 0.418 0.551 0.437
± ± ± ± ± ± ±
0.017 0.012 0.019 0.018 0.024 0.017 0.026
PCE [%]a 3.88 5.65 6.94 6.22 4.23 5.75 4.22
± ± ± ± ± ± ±
0.21 0.20 0.27 0.16 0.17 0.22 0.16
(4.10) (5.87) (7.24) (6.42) (4.41) (5.99) (4.39)
μe/μhb [cm2 V−1 s−1] 2.75c/5.66 10.3c/12.1 18.7c/16.3 11.6d/16.3 9.81e/16.3 / /
× × × × ×
10−5 10−5 10−5 10−5 10−5
a
Average values from 10 devices with the best PCE shown in brackets. bHole mobilities were measured with devices of ITO/PEDOT:PSS/active layer/Au. c,d,eElectron mobilities were measured with devices of ITO/TIPD/active layer/PDINO/Al, ITO/TIPD/active layer/Ca/Al, and ITO/ TIPD/active layer/Al, respectively.
EQE curves was 10.53, 10.37, and 8.69 mA/cm2 for the best solar cells of B, B+C, and C, respectively. Both hole and electron mobilities were estimated with the space-charge-limited-current (SCLC) method (Supporting Information and Figure S2), and Table 1 collects the values. The hole mobility decreased as the condition went from B to B +C and C, e.g., from SA only to SA+SVA and SVA only, which was consistent with the “dissolution” of white domains with the phase size decreasing from 250 to 30 and 25 nm (Figure S3a− c). We noted that the white phases could be contributed mainly from the polymer donor and the black phases from the PDI acceptor, according to our previously reported paper.13 The one-dimensional (1D) and two-dimensional (2D) grazingincidence X-ray diffraction (GIXRD)36 data (Figures 2c and S4) indicated close (010) diffraction patterns for all three blends in both the out-of-plane and in-plane directions. The more intense (010) diffraction along the out-of-plane than that along the in-plane indicated the preferable face-on orientation. As demonstrated in Figure 2c by the green lines, the (100) diffraction of the pure PTB7-Th film appeared as a broad band around 2θ = 3.8° and 4.4° along the out-of-plane direction and as a sharp band around 3.7° along the in-plane direction. As blended with the PDI acceptor, intense (100) diffractions were observed at 2θ = 4.0° along the out-of-plane and at 3.5° along the in-plane, each of which was positioned at a smaller 2θ value than that from the pure polymer film and were assumed to be mainly contributed from the acceptor. The different positions of the (100) diffractions at the out-of-plane and at the in-plane can be associated with the packing patterns of dimeric PDI molecules, for example, along the bay-to-bay and the nitrogento-nitrogen directions of perylene, respectively. The (100) diffraction intensity slightly decreased in both the out-of-plane and in-plane directions as the condition went from B to B+C
or SVA), the best cell shows four parameters as follows: Jsc = 9.17 ± 0.21 mA/cm2, Voc (open-circuit voltage) = 0.884 ± 0.008 V, FF = 0.430 ± 0.016, and the PCE = 3.47 ± 0.24% (hereafter, named as cell A, Table 1). Upon post treatment with SA in a fully covered Petri dish (hereafter, named as cell B) for 30 and 45 min, the Jsc increased to 10.89 ± 0.19 and 11.29 ± 0.13 mA/cm2, respectively. The Voc was of 0.874 ± 0.006 and 0.870 ± 0.008 V, and FF was of 0.453 ± 0.013 and 0.458 ± 0.017, respectively, which gave the PCE of 4.30 ± 0.25% and 4.46 ± 0.16% (Table S2). If 7.5 μL of the host processing solvent, 1,2-dichlorobenzene (o-DCB), was preadded into a fully covered Petri dish prior to the deposition of active layer (hereafter, named as cell C), the best cell was obtained after 15 min of SVA with the PCE = 4.60 ± 0.29%, Jsc = 10.03 ± 0.15 mA/cm2, Voc = 0.872 ± 0.009 V, and FF = 0.528 ± 0.014. SVA for a longer time led to a slight increase in FF and an obvious decrease in Jsc (Table S2). Comparisons of the SA and SVA results indicated that SA could give a higher Jsc, while SVA could be helpful for a higher FF. Therefore, we combined SA and SVA by first treating the as-cast solar cell blend in a fully covered Petri dish, for example, for 30 min (SA), and then the blend was quickly moved into another fully covered Petri dish with 7.5 μL of o-DCB inside, for example, for another 15 min (SVA) (hereafter, named as cell B + C). The optimization details were shown in Table S2. As a result, both the higher Jsc and FF values were maintained, giving the PCE value of 4.96 ± 0.24% with Jsc = 10.89 ± 0.22 mA/cm2, Voc = 0.876 ± 0.009 V, and FF = 0.523 ± 0.013 (Table 1). Figures 2a and 2b are the current density−voltage (J−V) characteristics and external quantum efficiency (EQE) responses of the best solar cells of B, C, and B+C, respectively, and Table 1 collects the average values of the four cell parameters obtained under conditions of A, B, C, and B+C, respectively. The current integrated from the 15417
DOI: 10.1021/acsami.6b03926 ACS Appl. Mater. Interfaces 2016, 8, 15415−15421
Research Article
ACS Applied Materials & Interfaces
Figure 3. Experimental J−V characteristics (a) and EQE response (b) of the optimal PTB7, PTB7-Th, and PBDT-TS1 based solar cells, respectively. The absorption spectra (c) and the out-of-plane (d) and in-plane (e) 1D GIXRD data of the corresponding solar cell blends.
Th, and PBDT-TS1 blends, respectively, which agreed well with the Jsc values of the devices. The cell’s Voc is thermodynamically determined by the energy difference between the donor’s HOMO and the acceptor’s LUMO, ED‑A. As the blended donor polymer was shifted from PTB7 to PTB7-Th and PBDT-TS1, respectively, the Voc increased from 0.840 ± 0.006 to 0.876 ± 0.007 and 0.887 ± 0.005 V, respectively, which was well consistent with the down trend of the HOMO energy, going from −5.02 eV to −5.17 eV and −5.26 eV (Figure 1c). The Jsc increased from 8.46 ± 0.19 to 11.14 ± 0.21 and 13.23 ± 0.19 mA/cm2, respectively, with the polymer shifting from PTB7 to PTB7-Th and PBDT-TS1. Red-shifting of the absorption edge of the polymer donor led to a wider spectral coverage of the solar emission (Figure 3c), contributing to the increase of Jsc. Again, the increase in both the hole and electron mobilities (Table 2) contributed to the increase of Jsc. Figures 3d and 3e as well as S6 are the 1D and 2D GIXRD data of the corresponding solar cell blends. All three polymers show preferable face-on orientation in either neat or blended films with stronger (010) diffraction intensity along the out-of-plane than along the in-plane direction. The (010) diffraction appeared at 2θ = 22.8° for PTB7, 22.6° for PTB7-Th, and 24.2° for PBDT-TS1 in their neat films. As blended with the PDI acceptor, the (010) diffraction of polymer shifts to 23.2°, 24.0°, and 24.6° for PTB7, PTB7-Th, and PBDT-TS1, respectively, corresponding to a ππ-stacking distance of 0.38, 0.37, and 0.36 nm, respectively. As we have observed, there were no obvious (010) diffractions seen from the PDI dimer, due to its highly twisted conformation.12 ππ-stacking of the polymer backbone becomes compacted, which was helpful for the hole transportation and contributed to the increase of the hole mobility. The out-of-plane (100) diffraction of the polymer in their neat films appeared at 4.8° for PTB7, 4.4° for PTB7-Th, and 3.4° for PBDT-TS1, respectively. For the three blended films, the (100) diffractions all appeared at a
and C. The electron mobilities were close to each other for the three cells with the estimated value slightly decreasing from cell B to B+C and C. The sum of the hole and electron mobility values decreased by the order of B > B+C > C, which agreed well with the decreased trend of the Jsc. Again the electron-tohole mobility became more and more balanced, contributing to the increase of the FF. The FF value is again related to the recombination losses of the mobile charges. Recombination losses of devices were reflected from incident light-power (P0) dependences of Jsc and Voc. The relationship between Jsc and light intensity follows the relation, Jsc ∝ Pα.37 For the three cells, the α values estimated from the plots of Jsc versus P0 (Figure S5a) were all close to each other, typically 0.95. At the open circuit, the recombination mechanism is reflected by the relation, Voc ∝ nkT/q ln(P), where k, T, and q are the Boltzmann constant, the temperature in Kelvin, and the elementary charge, respectively.37 The n values estimated from the plots of Voc versus P0 (Figure S5b) were close for cells B+C and C (1.08 vs 1.09 kT/ q), while that value from cell B was more deviated from one (1.31 kT/q), indicating stronger loss involved, which is again consistent with the lower FF value for cell B. By utilizing PDINO to replace the Ca layer, we again fabricated the solar cells under the condition B+C, the optimal solar cell showing an improved PCE (5.65 ± 0.20% vs 4.96 ± 0.24%) with an increasing FF (0.583 ± 0.012 vs 0.523 ± 0.013) and a close Jsc and Voc (Tables 2 and 1). A change of the polymer donor from PTB7-Th to a lower band gap polymer of PBDT-TS1 produced a higher PCE of 6.94 ± 0.27%, while changing to a wider band gap polymer of PTB7 led to a lower PCE of 3.88 ± 0.21% (Table 2). The optimization details were shown in Table S3. Figure 3a and b gives the J−V characteristics and EQE responses of the best solar cells from the PTB7, PTB7-Th, and PBDT-TS1 blends, respectively. The current integrated from the EQE curves was 8.17, 10.75, and 12.59 mA/cm2 for the best solar cells from the PTB7, PTB715418
DOI: 10.1021/acsami.6b03926 ACS Appl. Mater. Interfaces 2016, 8, 15415−15421
Research Article
ACS Applied Materials & Interfaces
Figure 4. Experimental J−V characteristics (a) and plots of Jsc and Voc versus the incident light power, P0, (b) of solar cells with different cathode layer, respectively.
indicating that the increase in Jsc with the cathode interlayer going from Ca to PDINO was contributed from the PDINO layer. We kindly note that a previous study on the PTB7:PC71BM system indicated the effects of methanol treatment on the bulk-heterojunction solar cell performance.38
close 2θ value of 4.0° along the out-of-plane. It was assumed to be mainly originated from the PDI dimer acceptor. TEM images (Figure S7) indicated that three solar cell blends all show nanoscale phase-separated black and white domains phase sizes all comparable to the exciton diffusion length. The estimated α value from PTB7 (0.938) is more deviated from one than that from the other two polymer-based solar cells (0.958 and 0.957) (Figure S8a), as does the n value (1.22 vs 1.05 and 1.07 kT/q) (Figure S8b). Both the α and n values suggest that stronger nongeminate recombination losses for the mobile charges were involved for PTB7, which well explained the lower FF value for PTB7-based solar cells (Table 2). Again, the hole and electron mobilities from the PTB7-based solar cell became more unbalanced (μh/μe = 2.06) than the other two cells (μh/μe = 1.17 and 0.87) (Table 2 and Figure S9). The photovoltaic results of the PTB7-Th:PDI system suggested that the electron-transporting layer (ETL) had an impact on the solar cell performance, due to the different electron mobilities: 9.01 × 10−5 for the Ca layer and 1.03 × 10−4 for the PDINO layer. Here, we selected PBDT-TS1:PDI combination to study the effects of ETL, and Table 2 collects the related photovoltaic results. Figures 4a and S10 display the J−V and EQE curves of the solar cells with Al, Ca/Al, and PDINO/Al as the cathode, respectively. The current integrated from the EQE curves was 11.38, 11.85, and 12.59 mA/cm2 for the best solar cells with Al, Ca/Al, and PDINO/Al as the cathode, respectively, in line with the Jsc values of the devices within 10% difference. Incorporation of the Ca layer helped dramatically reduce the bimolecular loss at the organic-to-metal interface, leading to an increase of FF and again Jsc. The n value was 1.32 kT/q and 1.08 kT/q for the Al and Ca/Al cells, respectively (Figure 4b), and the α value was 0.937 and 0.953, respectively. Both the n and α values indicated that weaker bimolecular recombination loss was involved in the Ca/Al cell. Reduction of the bimolecular loss contributes to a higher charge collection probability and, hence, a higher Jsc (Figure 4a). Both the Ca/Al and PDINO/Al cells show a close n value (1.08 kT/q vs 1.07 kT/q) and also a close α value (0.953 vs 0.957) (Figure 4b). However, the PDINO/Al cell showed a higher electron mobility than the Ca/Al cell (1.87 vs 1.16 × 10−4 cm2 V−1 s−1) (Table 2 and Figure S11). As comparisons, the Ca/Al and Al cells both show a close value of electron mobility. This suggests that the PDINO layer could afford an improved electron transportation than the Ca layer, which contributes to the higher electrical current generation and, hence, the higher Jsc and PCE. As shown in Table 2, processing the active layer with pure methanol only did not lead to an obvious change in Jsc and FF but a slight decrease in Voc,
■
CONCLUSIONS
■
EXPERIMENTAL SECTION
In summary, a high-efficiency PDI-based nonfullerene organic solar cell with a PCE of 6.94 ± 0.27% was achieved by optimizing both the hole and elctron mobilities via integration of film-morphology optimization, polymer donor selection (from PTB7 to PTB7-Th and PBDT-TS1), and cathode interlayer engineering. With PTB7-Th as the donor, filmmorphology optimizations using a combination of solvent annealing (SA) and solvent-vapor annealing (SVA) treatments led to a more balanced electron-to-hole mobility with respect to using the SA only and a higher electron and hole mobilities both with respect to using the SVA only. As a result, the combination of SA and SVA yielded a trade-off between FF and Jsc and, hence, a higher PCE of 5.65 ± 0.20%. A replacement of PTB7-Th with the lower band gap PBDT-TS1 afforded an improved PCE of 6.94 ± 0.27% due to the improved absorption and higher hole and electron mobilities. The cathode interlayer played an important role in solar cell’s performance: incorporation of Ca layer reduced bimolecular loss, giving a higher charge collection probability and a higher Jsc, while the PDINO layer afforded a higher electron mobility with respect to the Ca layer, leading to a higher electrical current generation.
Solar Cell Device Fabrications and Measurements. Indium tin oxide (ITO) coated glass substrates were precleaned with deionized water, CMOS grade acetone, and isopropanol in turn for 15 min. The organic residues were further removed by treating with UV-ozone for 1 h. Then they were coated with PEDOT:PSS (poly(3,4ethylenedioxythiophene):poly(styrenesulfonate)) (about 40 nm) at 2000 rpm for 35 s. After being baked at 150 °C for ∼15 min, the substrates were transferred into a nitrogen-filling glovebox (
