徐小云，吴宏波，梁世洁，唐正,＊，李梦阳，王静，王翔，闻瑾，周二军，李韦伟,＊，马在飞,＊

1东华大学材料科学与工程学院，纤维材料改性国家重点实验室，先进低维材料中心，上海 201620

2北京化工大学材料科学与工程学院，有机无机复合材料国家重点实验室，软物质科学与工程高精尖创新中心，北京 100029

3国家纳米科学中心，北京 100190

1 Introduction

Organic solar cells (OSCs) based on the bulk heterojunction(BHJ) concept have attracted more and more attention during the last two decades, owing to their advantages of low cost, light weight, high flexibility, and easy to manufacture1–3. The power conversion efficiency (PCE) of OSCs has been improved rapidly since the development of non-fullerene acceptor (NFA)materials4,5. Today, the highest PCE values reported for OSCs based on the NFA are over 18%6–8. Typically, the donor materials used in the state-of-the-art NFA OSCs are synthesized based on the donor-acceptor push-pull concept9,10. The most common building blocks for the donor materials are benzodithiophene (BDT) and benzodithiophenedione(BDD)11,12. For instance, poly[(2,6-(4,8-bis(5-(2-ethylhexyl)thiophen-2-yl)-benzo [1,2-b:4,5-b’]dithiophene))-alt-(5,5-(1’,3’-di-2-thienyl-5’,7’-bis(2-ethylhexyl)benzo[1’,2’-c:4’,5’-c’]dithiophene-4,8-dione) (PBDB-T), the model donor used in the first-generation NFA OSCs, has been reported to have desired energy levels, optical absorption, and miscibility for being used with the IT-series acceptors (such as 3,9-bis(2-methylene-(3-(1,1-dicyanomethylene)-indanone))-5,5,11,11-tetrakis(4-hexylphen-yl)-dithieno[2,3-d:2’,3’-d’]-s-indaceno[1,2-b:5,6-b’]dithiophe-ne (ITIC), 3,9-bis(2-methylene-((3-(1,1-dicyanome-thylene)-6,7-difluoro)-indanone))-5,5,11,11-tetrakis(4-hexylphenyl)-dithieno[2,3-d:2’,3’-d’]-s-indaceno[1,2-b:5,6-b’]dithiophene (IT4F), etc.)13. Later, the derivatives of PBDB-T, such as PM6, were developed, further improving the performance of the NFA OSCs14. However, these donor materials, requiring complicated synthetic procedures, could potentially lead to high production cost of OSCs, and hinder the industrialization of this emerging photovoltaic (PV)technology15,16.

Poly(3-hexylthiophene) (P3HT) was one of the most important donor materials in the history of organic PVs. It has a simple molecular structure, and thus, is easy and cheap to synthesize16–20, which are the paramount evaluation criteria for the scaling-up of OSCs. It has been reported that the synthesis of over 10 kg P3HT could be realized in a single batch, making it the most scalable donor material21. Besides, the batch-to-batch performance variation of P3HT was found to be very small22,23.In addition, owing to the large band gap of P3HT (allowing to realize high device output voltage), the P3HT OSCs based on NFA (with energy levels matching those of P3HT) are expected to be particularly attractive for indoor light harvesting, a unique advantage of OSCs (due to the superior performance of OSCs under weak illumination)24. This is because that the absorption of P3HT is restricted to the visible spectral range (400–700 nm).Ideally, for indoor applications, the photocurrent generation (a result of visible light excitation) in the P3HT-based OSCs could be as efficient as that in the solar cells based on PBDB-T or the derivatives of PBDB-T, but the upper limit for open-circuit voltage (Voc) is considerably higher for the P3HT solar cells25,26.

High PCE values (6.7%) for OSCs based on P3HT were first achieved using the fullerene derivative Indene-C60 Bisadduct(IC60BA) as the acceptor material27. However, due to the weak absorption of the fullerene acceptor, the short-circuit current density (Jsc) of these solar cells is limited. Besides, due to the misaligned energy levels of P3HT and IC60BA, voltage losses(Vloss) to the charge-transfer (CT) process are very high in the solar cells base on P3HT:IC60BA. This limits the Voc, and thus the overall performance of the solar cell28.

The development of NFAs, in principle, paves the way for the use of P3HT for highly efficient PV energy conversion, since absorption strength of NFAs is generally high, and it can be engineered to cover a wide range of the solar spectrum22,29.However, the performance of the P3HT solar cells based on the most commonly used NFA materials, including ITIC and 2,2’-((2Z,2’Z)-((12,13-bis(2-ethylhexyl)-3,9-diundecyl-12,13-dihydro[1,2,5]thiadiazolo[3,4-e]thieno[2ʺ,3ʺ:4ʹ,5ʹ]thieno[2ʹ,3ʹ:4,5]pyrrolo[3,2-g]thieno[2ʹ,3ʹ:4,5]thieno[3,2-b]indole-2,10-diyl)bis(methanylylidene))bis(5,6-difluoro-3-oxo-2,3-dihydro-1H-indene2,1-diylidene))dimalononitrile(Y6), is very poor, with PCE generally lower than 2%29–31. This was not only due to the high Vloss, associated with the misaligned energy levels, but also related to the low Jsccaused by high quantum efficiency (QE)losses in the device. In addition, the low external quantum efficiency of electroluminescence (EQEEL), causing high nonradiative recombination voltage losses (∆Vnr), further limits the Vocof the P3HT:NFA OSCs22,29,31,32. Although additional acceptor materials, such as IDTBR33, BTA126, and TrBTIC23,have been developed for the solar cells based on P3HT, allowing to achieve increased device QE, the overall performance of the P3HT based OSCs is still much worse, as compared to that of the state-of-the-art OSCs, due to the higher Vloss6,9,34.

Recently, it has been demonstrated that the poor performance of the P3HT based OSCs could be ascribed to the high miscibility between P3HT and NFA, leading to too-small phase separation between the donor and the acceptor molecules in the photoactive layer29,31,35. The too-small phase separation could lead to fast recombination loss of charge carriers, thus limit both the Jscand the fill factor (FF) of the solar cell31,36. However, a too-small phase separation should increase both the radiative(Kr) and the non-radiative decay rate (Knr) of charge carriers,thus, have limited impact on the EQEEL, and thus, the ∆Vnrof the solar cell, since37,38

Therefore, the generally lower EQEEL, the higher ∆Vnr, and the limited Vocin the P3HT:NFA OSCs, as compared to those in the state-of-the-art OSCs, could not be ascribed to the too-small phase separation. An additional reason for the limited Voc, and thus, the limited PV performance of the P3HT:NFA OSCs must exist.

To understand the reason for the limited performance of the P3HT:NFA based OSCs, and for the development of an effective strategy to improve the device performance, in this work, we investigate P3HT solar cells based on fullerene and the typical non-fullerene acceptors, including IT4F9and Y639. We find that the photovoltaic QE of the P3HT:NFA solar cells is much lower,compared to that of the P3HT:fullerene solar cell, due to the higher decay rate of charge carriers (K), and the origin of the higher K is found to be the higher Knr. The higher Knralso leads to significantly higher ∆Vnrin the P3HT:NFA solar cells, limiting the device Voc. Since the energetic properties of the NFAs (IT4F,Y6) are not very different from that of the fullerene acceptor, the most likely reason for the increased Knrof the P3HT:NFA solar cells is ascribed to the too-short molecular distance between P3HT and the NFA in the blend active layer. Then, we investigate the P3HT solar cell based on the NFA 2,2ʹ-((12,13-bis(2-butyldecyl)-3,9-diundecyl-12,13-dihydro-[1,2,5]-thiadiazolo[3,4-e]thieno[2ʺ,3ʺ:4ʹ,5ʹ]thieno[2ʹ,3ʹ:4,5]p-yrolo[3,2-g]thieno [2ʹ,3ʹ:4,5]thieno[3,2-b]indole-2,10-diyl)bis(methanely-lidene))bis(5,6-dichloro-1H-indene-1,3(2H)-dione) (ZY-4Cl), with a chemical structure similar to Y6. The performance of the P3HT:ZY-4Cl solar cell is found to be much better than that based on P3HT:Y6 or P3HT: fullerene, mainly due to the reduced Knr, associated with the better alignment of energy levels in the blend of P3HT:ZY-4Cl. In addition, the donor-acceptor distance in the blend of P3HT:ZY-4Cl is longer, contributing additionally to the reduced Knr. Nevertheless, the performance of the solar cell based on P3HT:ZY-4Cl is still limited by the Knr. Thus, we propose that to reduce Knrand improve the performance of P3HT based OSCs, the energetics of the CT states should be further optimized, and the distance between the donor and the acceptor molecules (DA distance) in the BHJ active layers should be increased.

2 Experimental and computational section

2.1 Materials

P3HT was purchased from Sigma Aldrich. PC61BM was purchased from Solenne BV. Y6, IT4F and ZY-4Cl were purchased from Solarmer Materials (Beijing, China). All the reagents used in this work were purchased from Sigma Aldrich.

2.2 Device fabrication

The BHJ OSCs based on P3HT mixed with different acceptors were fabricated with an inverted device architecture of glass/ITO/ZnO (30 nm)/active layer/MoO3(10 nm)/Ag (100 nm). The pre-patterned ITO substrates were first cleaned by detergent, then by TL1 solution (a mixture of NH3·H2O (25%) :H2O2(30%): ultra-pure water in volume ratio of 1 : 1 : 5, 60 °C).The ZnO interlayers were deposited using a sol-gel method: The sol-gel precursor solution was prepared by mixing zinc acetate dihydrate (Aldrich, 99.9%, 1 g), ethanolamine (Aldrich, 99.5%,277 µL) and dimethoxy ethanol (Aldrich, 99.8%, 10 mL). The precursor solution was spin-coated (4000 r·min−1) on top of the clean ITO substrates and the substrates were subsequently annealed at 200 °C for 30 min on a hot plate. The active layer solutions were prepared by dissolving the P3HT:PC61BM,P3HT:Y6 and P3HT:IT4F in o-dichlorobenzene (oDCB) with a donor : acceptor (D : A) weight ratio of 1 : 1. The total concentrations of the solutions were 35, 22 and 22 mg·mL−1,respectively. The active layer solutions were heated at 85 °C and stirred at a speed of 1000 r·min−1for 12 h, prior to use. For the P3HT:ZY-4Cl based solar cells, the active solution had a total concentration of 14 mg·mL−1, tetrahydrofuran (THF) was used as the solvent, and the D : A weight ratio was 1 : 1. The thickness of the active layer based on P3HT:PC61BM was 200 nm, and the thickness of the active layers based on P3HT:non-fullerene was 100 nm. The active layers were spin-coated on top of the ZnO coated substrates. Then it was annealed on a hot plate for a certain temperature. For solvent vapor annealing (SVA)treatment on the active layers is performed in an atmosphere of solvent oDCB for 12 h. The substrates coated with active layers were then transferred to a vacuum chamber mounted in a glove box filled with nitrogen. Subsequently, 10 nm MoO3and 100 nm Ag were thermally deposited through a shadow mask onto the active layers under a vacuum pressure of 10−4Pa. The photoactive area for these OSC devices was 0.04 cm2.

Details regarding the characterization of the devices are provided in Supporting Information.

3 Results and discussion

3.1 Quantum efficiency losses in solar cells based on P3HT:NFA

To evaluate the difference in the PV performance of the OSCs based on P3HT:fullerene and P3HT:NFA, solar cells with an inverted architecture of glass/ITO/ZnO (30 nm)/active layer/MoO3(10 nm)/Ag (100 nm) were constructed. The BHJ active layer systems are based on P3HT:PC61BM, P3HT:Y6, and P3HT:IT4F (chemical structures shown in Fig. S1, Supporting Information), and the energy levels (the lowest unoccupied molecular orbital (LUMO) and highest occupied molecular orbital (HOMO) levels) of the active materials are provided in Fig. 1a40,41. The absorption spectra of the active blends are given in Fig. S2 (Supporting Information). The PV performance of the solar cells is optimized by thermal annealing and SVA methods.The performance for the optimized solar cells was characterized using a standard solar simulator (AM1.5 G) with an illumination intensity of 100 mW·cm−2, and the current density–voltage (J–V) curves of the optimized solar cells are plotted in Fig. 1b(details regarding the optimization of the solar cells are provided in SI-4, Supporting Information). Their corresponding external quantum efficiency (EQE) spectra are given in Fig. 1c.

From the J–V curves, we note that Jscs of the solar cells based on fullerene and NFA are remarkably different. Despite having improved optical absorption of the active layer (Fig. S2,Supporting Information), the solar cell based on P3HT:IT4F has much worse Jsc(5.08 mA·cm−2), compared to that based on P3HT:PC61BM (8.93 mA·cm−2). In addition, the Jscof the solar cell based on P3HT:Y6 is 10.77 mA·cm−2, which is higher than that based on P3HT:PC61BM, but much lower than one would expect for an OSC based on Y6 with the absorption edge at over 900 nm (for instance, Jscof the solar cell based on PM6:Y6 is over 25 mA·cm−2)39. The severely limited Jscs of the P3HT OSCs based on the Y6 and IT4F suggest that the internal quantum efficiencies (IQEs) of the NFA devices are very low. Therefore,transfer matrix model (TMM) simulations were carried out using the real dielectric constants of the materials42,43, to estimate the IQEs of the P3HT OSCs: The IQE spectra of the solar cells,calculated by dividing the measured EQE spectra by the TMM simulated EQE spectra, are shown in Fig. 1c, and the maximum short-circuit current density (Jsc-max) values, predicted by TMM(IQE = 100%), are plotted as a function of the active layer thicknesses for the solar cells, as shown in Fig. 1d. From Fig. 1c and 1d, we find that the IQEs of the OSCs based on P3HT:Y6 (≈55%) and P3HT:IT4F (≈ 30%) are indeed much lower than that of the P3HT:PC61BM based solar cell (≈ 75%).

Fig. 1 (a) HOMO and LUMO energy levels of P3HT, PC61BM, Y6, and IT4F. (b) J–V curves, (c) EQE and IQE spectra of the P3HT based OSCs.(d) Jsc-max as a function of the active layer thickness predicted by TMM simulations for the P3HT based OSCs.

The misaligned energy levels of the donor and the acceptor molecules for the P3HT:NFA system are expected to limit the performance of the NFA solar cells. However, it should not limit the IQE of the device. To understand the reason for the limited IQE, we divide IQE into three parts44:

where ηextis the dissociation efficiency of the singlet excitons generated in the pristine donor or acceptor phases in the active layer, ηCTis the dissociation efficiency of the CT excitons formed at the donor/acceptor (D/A) interfaces, and ηFCis the collection efficiency of free charge carriers at the electrodes. The lower IQE could be a result of low ηext, which is mainly determined by the degree of phase separation between the donor and the acceptor materials in the BHJ systems. To evaluate the degree of phase separation in the active layer blends based on P3HT, atomic force microscopy (AFM) measurements were performed, and the results are given in Fig. 2a. The AFM images show that the topographic properties of the three blends are similar. The root-mean-square (RMS) surface roughness values are 3.38, 3.53, and 2.61 nm for the P3HT:PC61BM, P3HT:Y6,and P3HT:IT4F active layer blends, respectively, which suggests that the degree of phase separation in the active layer based on P3HT is low, regardless of the acceptor used. Photoluminescence(PL) measurements were also performed to evaluate the morphological properties of the active layers based on P3HT. As shown in Fig. 2b, the PL spectra reveal that the emission of P3HT is significantly quenched in the active layer based on either the fullerene or the non-fullerene acceptors, indicating that the exciton dissociation efficiency is high in all of the blends. Thus,the limited IQE of the OSCs based on P3HT:Y6 and P3HT:IT4F could not be ascribed to low ηext. Also, ηCTis expected to be high for all of the blends, since the energetic driving force for exciton dissociation40is sufficiently large (Fig. 1a). Thus, ηCTshould not be the main reason for the limited IQE of the P3HT:NFA solar cells. Therefore, we expect that the extraction efficiency of free charge carriers, ηFC, must be very low for the P3HT:NFA solar cells, being the reason for the limited device IQE.

For OSCs, extraction efficiency of charge carriers depends on the recombination rate of charge carriers. Therefore, light intensity dependent J–V measurements were performed for the OSCs based on P3HT:PC61BM, P3HT:Y6, and P3HT:IT4F: We find that the slopes of the Voc-light intensity curves (Fig. 2c) are about 1.5 kT for both the OSCs based on P3HT:fullerene and P3HT:NFA, suggesting that bimolecular recombination exists in these solar cells31,45. Also, the Jscof the OSCs are plotted as a function of light intensity, as shown in Fig. 2c. The dependence of Jscon light intensity for the solar cell based on P3HT:PC61BM is found to be strictly linear (with a slope of 0.96), suggesting that the bimolecular recombination is not limiting the IQE of the solar cell under short circuit. However, the dependence of Jscon light intensity is sub-linear, with a slope of 0.89 and 0.75, for the solar cells based on P3HT:Y6 and P3HT:IT4F, respectively,suggesting that the Jscs of the solar cells are strongly limited by the bimolecular recombination. These results indicate that the IQE, and thus, the Jscof the P3HT:NFA OSCs is indeed limited by the extraction efficiency of free charge carriers.

Fig. 2 (a) AFM height images (size 5 μm × 5 μm) of the P3HT-based blend films. (b) PL spectra of the pure P3HT film and the P3HT-based blend films, measured with laser excitation at 550 nm. Quenching efficiency values are indicated in the plots. (c) Dependence of Voc and dependence of Jsc on light intensity for the OSCs based on P3HT:fullerene and P3HT:NFA, determined from the light intensity dependent J–V measurements.

To investigate the reason for the severe bimolecular recombination losses of IQE in the solar cells based on P3HT:NFA, transient photovoltage decay (TPV) measurements were carried out. The transient voltage decay signals,representing the decay dynamics of free charge carriers, are measured at different bias illumination intensities (leading to different bias photovoltage, Vph), as shown in Fig. 3a. Details regarding the TPV measurements are provided in SI-1(Supporting Information). By fitting the voltage decay signals with an exponential decay function (Fig. 3a), we derived the voltage decay lifetimes (related to the lifetime of the free charge carriers) as a function of Vph. As shown in Fig. 3b, we find that the transient voltage decay lifetimes of the solar cells based on P3HT:Y6 and P3HT:IT4F are indeed much shorter than that of the P3HT:PC61BM solar cell. Thus, the bimolecular recombination rate (K), inversely proportional to the lifetime of charge carriers, is much higher for the P3HT:NFA solar cells than that for the fullerene solar cell, limiting the extraction efficiency of free charge carriers and IQE of the solar cells based on P3HT:NFA.

3.2 High Vloss in OSCs based on P3HT:NFA

The high overall recombination rate, K, i.e., the sum of Krand Knr, of the OSCs based on P3HT:NFA could be a result of the very small phase separation between the donor and the acceptor materials in the active layer, as is pointed out in the literature29,31,35.However, as discussed above, the small phase separation should lead to high Krand high Knr. Therefore, EQEEL(= Kr/Knr), and thus, ∆Vnrof the solar cell would not necessarily be limited by the high K.

To determine ∆Vnrof the P3HT based OSCs, the radiative recombination limit for the Voc(Voc,rad) is calculated from the highly sensitive EQE (sEQE) spectra (Fig. 3c), using Equations(3) and (4)46:

Fig. 3 (a) TPV decay signals and (b) TPV decay time as a function of Vph (generated by the bias illumination) derived from the TPV measurements for the P3HT based OSCs. (c) sEQE spectra of the P3HT based OSCs and the spectra calculated from the product of EQE(E) and BB(E)(for the determination of J0,rad and Voc,rad). The dashed lines represent the upper and the lower limits for the fitting curves.Details regarding the determination of ECT are provided in SI-2 (Supporting Information).

where k is the Boltzmann constant, BB(E) is the blackbody emission photon flux, J0,radis the radiative recombination limit for the dark saturation current density, and Jphis the photocurrent density generated in the solar cell under an open-circuit condition (assumed to be equivalent to Jsc). Then, ∆Vnrof the solar cells are determined, using

We note that ∆Vnrin the solar cells based on P3HT:Y6 (0.56 V) and P3HT:IT4F (0.54 V) are significantly higher than that of the P3HT:PC61BM solar cell (0.37 V), as listed in Table 1.Therefore, a small phase separation should not be the only reason for the higher K in the solar cell based on P3HT:NFA, as compared to that based on P3HT:PC61BM.

We also constructed the P3HT:Y6 OSCs with different D : A weight ratios, and thus different degrees of phase separation between the donor and the acceptor materials31,37,47. The PV performance parameters for these solar cells are listed in Table S4 (Supporting Information). We find that ∆Vnr, K, as well as Vocof the solar cells based on P3HT:Y6 are hardly affected by the change of the D : A ratio (Table S4, Supporting Information).This further confirms that the high K in the P3HT:NFA based OSCs is not solely due to the too-small phase separation.

Table 1 The representative PV performance parameters and Vloss values for the P3HT-based OSCs.

We now investigate the reason for the high ∆Vnrin the solar cells based on P3HT:NFA. In organic solar cells, a high ∆Vnrvalue could either be due to a low Kror a high Knr. Kris related to the radiative voltage losses (∆Vr) of the device, via the following equation47:

where ∆Vrrepresents the Vlossin the ideal OSC without any nonradiative recombination voltage losses, and ∆Vrcan be expressed as46:

where ECTis assumed to be the effective bandgap of the BHJ active layers, which can be determined by a fitting to the lower energy part of the sEQE spectrum. Using the method described in the literature48, we find that ECTs of these solar cells are similar (Table 1), which is expected, since the HOMO(donor)-LUMO(acceptor) energetic offset of P3HT:fullerene is similar to that of the P3HT:NFA systems studied here. Therefore, ∆Vrvalues are found to be similar for these solar cells based on fullerene or NFA. Accordingly, Krof the P3HT:NFA OSCs are not expected to be much lower than that of the P3HT:fullerene OSC. As a result, the higher ∆Vnrof the P3HT:NFA OSCs has to be due to the higher Knr.

For BHJ OSCs, Knrcould be expressed as49:

where V is the electronic coupling matrix element, ħ is the reduced Planck’s constant, λ is the reorganization energy of CT state. According to Equation (8), Knrexponentially increases with decreasing of ECTand increasing of λ. However, as listed in Table 1, ECTof the OSCs based on P3HT:Y6, P3HT:IT4F, and P3HT:PC61BM are similar, about 1.15–1.18 eV. Also, the λ values of the solar cells determined from the sEQE spectra, using the method described in the literature, are similar (0.19–0.26 eV)48.Therefore, the increased Knrin the solar cells based on P3HT:NFA, as compared to that based on P3HT:fullerene, could not be explained by the change of ECTor λ (Knrof the solar cells based on P3HT:NFA, according to the EQEELand TPV results,should be approximately 2 orders of magnitude higher than that based on P3HT:fullerene).

V is another critical parameter determining Knr, according to Equation (8). In the generalized Mulliken-Hush (GMH) theory,V increases with increasing transition dipole moment (M) and reducing static dipole moment (µCT) of the CT state50,51:

M is proportional to the absorption oscillator strength of CT state (fosc), which exponentially increases with the reducing distance between the donor and the acceptor molecules forming the CT state (DA distance)52,53. µCTalso depends on the DA distance: µCTreduces with reducing DA distance51,52. Therefore,the DA distance plays an extremely important role in determining the Knr, and thus, the overall voltage losses in OSCs.This has recently been demonstrated for organic solar cells based on the single component active layers, as well as the highefficiency BHJ active layers54. Thus, we can only expect that the shorter DA distance, leading to larger V, is the reason for the higher Knrof the P3HT:NFA solar cells, compared to that of the P3HT:fullerene solar cell.

3.3 Reducing Knr in the P3HT:NFA OSCs

Recently, it has been demonstrated that the PCE of the P3HT based OSCs could be significantly increased using ZY-4Cl, with the chemical structure very similar to that of Y6 (Fig. S1,Supporting Information), as the acceptor material (PCE ≈10.24%)55. This is one of the highest PCE values realized for the OSCs based on P3HT22,23,26,33. To understand the reason for the improved device performance, solar cells based on P3HT:ZY-4Cl and P3HT:Y6 were also constructed in this work, using exactly the same processing conditions. The J–V curve and the EQE spectrum of the P3HT:ZY-4Cl OSC are shown in Fig. 4a and 4b, respectively. We note that the Jsc(14.78 mA·cm−2) in the solar cell based on P3HT:ZY-4Cl is indeed much higher,compared to that based on P3HT:Y6 (10.77 mA·cm−2), despite that the absorption band of ZY-4Cl is much narrower than that of Y6 (Fig. S1, Supporting Information). Thus, the higher Jscof the P3HT:ZY-4Cl based solar cell is mainly due to the higher device IQE. This is confirmed by the TMM simulations, which predict that the IQE of the solar cell based on P3HT:ZY-4Cl is about 80%, as shown in Fig. 4b. Thus, the loss of QE due to nonradiative recombination of charge carriers is significantly reduced in the P3HT:ZY-4Cl based OSC. TPV measurements(Fig. 4c) reveal that the lifetime of free charge carriers in the solar cell based on P3HT:ZY-4Cl is much longer, and thus, Knr(≈ K, for Knr>> Kr) is much lower than that based on P3HT:Y6.The lower Knrleads to significantly improved IQE, as well as reduced ∆Vnr(0.33 V), and thus, could give rise to the higher Jscand higher Voc(0.85 V) for the solar cell based on P3HT:ZY-4Cl.

Fig. 4 (a) J–V curves and (b) EQE and IQE spectra of the P3HT:ZY-4Cl and P3HT:Y6 based OSCs. (c) Voltage decay lifetimes of the solar cells based on P3HT:ZY-4Cl, P3HT:Y6, and P3HT:PC61BM, derived from the TPV measurements. (d) sEQE spectrum of the solar cell based on P3HT:ZY-4Cl, and the spectrum calculated from the product of EQE and BB (for the determination of J0,rad and Voc,rad).

To understand the reason for the reduced Knrof the solar cell based on ZY-4Cl, ECTof the solar cell is measured, as shown in Fig. 4d. ECTof the P3HT:ZY-4Cl solar cell is 1.41 eV, much higher than that of the solar cell based on Y6 (1.17 eV). The increase of the ECTvalue is expected because of the better alignment of the donor-acceptor energy levels of the P3HT:ZY-4Cl system (the LUMO and HOMO levels of ZY-4Cl are at−3.67 and −5.64 eV, respectively), compared to that of the P3HT:Y6 system. Thus, we expect that the increased ECTis the main reason for the reduced Knrin the P3HT:ZY-4Cl based OSC.

To evaluate whether the lower Knrof the solar cell based P3HT:ZY-4Cl could also be associated with the increased DA distance, we performed molecular dynamic (MD) simulations(Fig. 5) for the solar cells based on P3HT:ZY-4Cl and P3HT:Y6.Details regarding the MD simulations are given in SI-1(Supporting Information). From the MD simulations, we noted that the shortest DA distances in both the P3HT:ZY-4Cl and the P3HT:Y6 systems are dominated by the π–π stacking of the donor and acceptor molecules. Using the centroids to represent the locations of the conjugated groups in the donor and the acceptor molecules, we derive the radical distribution function(RDF) for the shortest DA distances, and we note that the shortest DA distances of the blend of P3HT:ZY-4Cl peaked at about 0.45 nm, longer than that of the P3HT:Y6 blend (0.41 nm).Therefore, the longer DA distance, leading to a lower V value, is expected to contribute additionally to the lower Knrof the ZY-4Cl based solar cell.

Fig. 5 Molecular dynamic simulation results for the blend of (a) P3HT:Y6 and (b) P3HT:ZY-4Cl. (c) Radical distribution functions for the shortest distances between the donor and the acceptor molecules in the blends of P3HT:ZY-4Cl and P3HT:Y6.

However, the Knrof the P3HT:ZY-4Cl solar cell is still high,leading to the ∆Vnrof 0.33 V, much higher than that of state-ofthe-art NFA OSCs (less than 0.2 V)46,49,56. The high Knrcould also be the reason for the limited IQE (≈ 80%) of the P3HT:ZY-4Cl solar cell. Therefore, the performance of the P3HT:ZY-4Cl solar cell is still limited. Considering the fact that the ECTof the P3HT:ZY-4Cl solar cell (1.41 eV) is already comparable to that of the most efficient organic solar cells, e.g., the PM6:Y6 solar cells, we expect that the DA distance is still too short in the blend of P3HT:ZY-4Cl, being the reason for the high Knrof the ZY-4Cl based solar cell.

It should be noted that the value for the DA distance derived from the MD simulation depends on the chemical structure used for the determination of the centroids of the donor-acceptor material. In this work, we only compare the P3HT systems based on the acceptors with similar chemical structures (Y6 vs. ZY-4Cl), and we used the centroids of the thiophene units in the P3HT molecule for the determination of the location of the donor at the DA interface. However, it is unclear how much shorter the DA distance in the P3HT:ZY-4Cl solar cells is, as compared to that in the state-of-the-art OSCs. Since the ECTof the solar cell based on P3HT:ZY-4Cl (1.41 eV) is already rather optimal for an OSC, furthering increasing ECT(for the purpose of reducing Knr) is expected to lead to limited spectral response of the device,thus, would not be beneficial for the device performance.Therefore, we expect that further increasing the DA distance would be the more desired strategy to reduce Knrand improve the performance of the P3HT:NFA based OSC.

4 Conclusions

To conclude, we investigated the reason for the low QE and high Vlossin the OSCs based on P3HT mixed with NFA. We found that the QE of the P3HT:Y6 and P3HT:IT4F based OSCs were considerably lower than that of the P3HT:PC61BM OSC,due to the higher charge carrier recombination rate K, in particular, the higher Knrof the CT states. This is a newly discovered origin for the limited PV performance in the P3HT:NFA based OSCs. It also explained the high ∆Vnrin the solar cells based on P3HT:Y6 and P3HT:IT4F. Also, we demonstrated that the use of ZY-4Cl as the acceptor material could lead to the better alignment of the energy levels in the active layer, and thus reduced Knr. In addition, the DA distance in the blend of P3HT:ZY-4Cl was found to be larger than that of the P3HT:Y6, further reducing the Knr, which allowed for the realization of the considerably improved solar cell performance.Nevertheless, the Knrin the P3HT:ZY-4Cl solar cell is still high,limiting the device Voc. To further improve the performance of the P3HT solar cells, the Knrshould be reduced, which could be realized by increasing the spacing between the donor and acceptor molecules in the P3HT:NFA active layer. This could be achieved by adjusting the composition of the active layer,extending the side chain length, or engineering the nonconjugated part of the active materials.

Supporting Information:available free of charge via the internet at http://www.whxb.pku.edu.cn.