Solvent control of the morphology of the hole transport layer for high-performance perovskite solar cells

Chemical Physics Letters 687 (2017) 258–263

Contents lists available at ScienceDirect

Chemical Physics Letters journal homepage: www.elsevier.com/locate/cplett

Research paper

Solvent control of the morphology of the hole transport layer for highperformance perovskite solar cells Xiaoyin Xie a, Guanchen Liu b, Li Chen a, Shuangcui Li c, Zhihai Liu a,d,e,⇑ a

Department of Chemical Technology, Jilin Institute of Chemical Technology, Jilin 132022, China Department of Material Science and Technology, Jilin Institute of Chemical Technology, Jilin 132022, China c Yantai Automobile Engineering Professional College, Shandong 264000, China d Department of Bio-Nano Technology, Gachon University, Gyeonggi 461-701, Republic of Korea e Gachon Bio-Nano Research Institute, Gyeonggi 461-701, Republic of Korea b

a r t i c l e

i n f o

Article history: Received 16 July 2017 In final form 15 September 2017 Available online 18 September 2017 Keywords: Solvent Hole transport layer Morphology Perovskite solar cells High performance

a b s t r a c t We investigated the effect of the morphology of 2,20 ,7,70 -tetrakis-(N,N-di-p-methoxyphenylamine)-9,90 spirobifluorene (spiro-OMeTAD) prepared using chlorobenzene (CB) and 1,2-dichlorobenzene (DCB) on the performance of perovskite solar cells (PSCs). We find that a more uniform and smoother spiroOMeTAD layer was obtained using DCB than CB. The PSCs prepared using DCB exhibited a higher power conversion efficiency (PCE = 16.2%) than those obtained using CB (PCE = 14.5%). The hysteresis was reduced from 4.8% to 0.6%, with improved stability. The highest PCE of PSCs prepared using DCB was 16.6%, indicating that the use of DCB for spiro-OMeTAD processing enables the fabrication of highperformance PSCs. Ó 2017 Elsevier B.V. All rights reserved.

1. Introduction Following the first application of methylammonium lead halide perovskites (CH3NH3PbX3, X = Cl, Br, or I) in solar cells by Miyasaka and coworkers in 2009 [1], the development of perovskite solar cells (PSCs) has been very rapid, because of their excellent performances and simple fabrication process. A record power conversion efficiency (PCE) of 22.1% was recently reported for a single PSC [2], highlighting the great potential of these systems for future commercialization [2–6]. Common PSCs have a typical p-i-n structure, with the perovskite sandwiched between the hole and electron transport layers [3–7]. A mesoporous TiO2 layer is usually deposited onto a compact thin TiO2 layer, fabricated on a fluorine-doped tin oxide (FTO)-coated glass substrate, to form the n-type electron transport layer [3–7]. In the case of the hole transport layer, 2,20 ,7,70 -tetrakis(N,N-di-p-methoxyphenylamine)-9,90 -spirobifluorene (spiroOMeTAD) is widely used for extracting holes from the perovskite layers and transporting them to the metal electrode [3–7]. Besides the specific materials used as perovskite absorbers and electron/hole transporters, the morphology of each layer plays a very ⇑ Corresponding author at: Gachon University, Department of Bio-Nano Technology, 1342 SeongnamDaero, Sujeong-Gu, Seongnam, Gyeonggi 461-701, Republic of Korea. E-mail address: [email protected] (Z. Liu). https://doi.org/10.1016/j.cplett.2017.09.031 0009-2614/Ó 2017 Elsevier B.V. All rights reserved.

important role in the device performance [8–17]. For examples, the morphology of the perovskite layer may affect its light harvesting and charge generation capabilities [8], whereas the morphology of the electron and hole transport layers has a strong influence on the charge transport, dissociation, and collection processes [9–17]. Recent investigations of the effect of the TiO2 morphology on the performance of PSCs have led to the use of a mesoporous TiO2 layer, a compact TiO2 layer, and TiO2 nanomaterials of different size as the electron transport layer in highperformance PSCs [3–7,10,13–15]. On the other hand, in the case of the hole transport layer, doping with organic and/or inorganic materials is a widely used strategy to improve the hole transport properties of spiro-OMeTAD [3–7,16,17]. The preparation of the spiro-OMeTAD layer usually involves dissolving spiro-OMeTAD in chlorobenzene (CB) [3–7,16,17]. In addition, ethyl acetate has been used for spiro-OMeTAD preparation, which resulted in a high PCE of 19.4% [18]. However, other solvents have not been widely used for spiro-OMeTAD layer preparation. Furthermore, the control of the morphology of the spiro-OMeTAD layer has not been thoroughly investigated so far. Solvent engineering is a widely used strategy to control the morphology of organic layers [11,12], which highlights the importance of investigating the effect of solventrelated changes in the morphology of the spiro-OMeTAD layer on the performance of PSCs. In this work, we investigate the effect of the spiro-OMeTAD morphology on the performance of PSCs by using CB and

X. Xie et al. / Chemical Physics Letters 687 (2017) 258–263

1,2-dichlorobenzene (DCB) for the deposition of the spiro-OMeTAD layer. The analysis shows that DCB leads to a smoother spiroOMeTAD layer compared with CB. This finding is attributed to the high boiling point and low vapor pressure of DCB, which considerably reduces the solvent evaporation rate during the formation of the spiro-OMeTAD layer by solidification. The improved morphology of the spiro-OMeTAD layer enhances its contact with the perovskite and the metal electrode. As a result, the PCE of PSCs based on a DCB-processed spiro-OMeTAD layer increased from 14.5% to 16.2%, with a simultaneous enhancement in the open circuit voltage (Voc), short circuit current density (Jsc), and fill factor (FF) characteristics. Furthermore, the DCB-processed spiroOMeTAD layer led to significantly reduced hysteresis (from 4.8% to 0.6%) and improved stability. The best PSC prepared using the DCB-processed spiro-OMeTAD layer showed a PCE as high as 16.6%. The present results indicate that using DCB for processing spiro-OMeTAD represents a simple and effective strategy for fabricating high-efficiency, low-hysteresis, and stable PSCs.

259

tonate) (Sigma-Aldrich, USA) in 1-butanol solution. Then, a commercial TiO2 paste (18NRT, Dyesol) diluted in ethanol was spincoated on the compact TiO2 layer. The TiO2-coated substrates were annealed at 500 °C for 30 min, following which the CH3NH3PbI3 perovskite precursor solution was spin-coated on the FTO/TiO2 substrate at 5000 rpm in a N2-filled glove box. During the spincoating process, 120 lL CB was quickly added on the surface of the substrate after a specific delay time of 6 s to form a smooth and pinhole-free perovskite layer. The hole transport material was prepared by dissolving 75 mg mL1 spiro-OMeTAD, 28 lL 4tert-butylpyridine, and 18 lL of a solution of bis(trifluoromethane) sulfonamide lithium salt (520 mg in 1 mL acetonitrile) in 1 mL CB or DCB. Then, the solutions were separately spin-cast onto the perovskite film at 3000 rpm for 30 s. Finally, an Au anode of 100 nm was deposited on the devices under a vacuum of 104 Pa. The effective working area of the PSCs, determined by a shadow mask, was 0.1 cm2. 2.2. Characterization

2. Experimental section 2.1. Device fabrication Lead iodide (PbI2), spiro-OMeTAD, and methylammonium iodide (MAI) were purchased from Sigma-Aldrich (USA), Nano-C (USA), and Xi’an Polymer Light Technology Corp. (China), respectively. All the solvents (CB, DCB, ethanol, 1-butanol, isopropanol, and N,N-dimethylformamide) used in this work were purchased from Sigma-Aldrich (USA). The perovskite precursor solution was prepared by dissolving MAI and PbI2 (1:1 M ratio) in anhydrous N,N-dimethylformamide with a total concentration of 45 wt%. As shown in Fig. 1(a), the PSCs were fabricated in a FTO/TiO2/ perovskite/spiro-OMeTAD/Au configuration. First, a 20 nm-thick compact TiO2 layer was coated on the cleaned FTO substrates by aerosol spray deposition of titanium diisopropoxide bis(acetylace-

The surface morphology of the perovskite and spiro-OMeTAD layers was inspected by atomic force microscopy (AFM, Veeco, USA). Ultraviolet–visible (UV–vis) absorption spectra were measured by a Lambda 750 (Perkin Elmer, USA) spectrometer. The photoluminescence (PL) spectra were measured using a spectrometer (FLS920, Edinburgh Instruments, UK). Cross-section scanning electron microscopy (SEM) images were acquired using a microscope (Hatachi, Japan) operated at an acceleration voltage of 12 kV. Electrochemical impedance spectroscopy (EIS) analysis was performed in dark conditions using a SP-240 potentiostat (Bio-Logic, France) in the frequency range of 0.1 Hz to 7 MHz. The current density– voltage (J–V) characteristics of the PSCs were measured under an irradiation intensity of 100 mW cm2 (1 sun, AM 1.5). The incident photon-to-current efficiency (IPCE) was measured using a Solar Cell IPCE measurement system (Solar Cell Scan 100, Zolix, China).

Fig. 1. (a) Schematic illustration of the PSCs investigated in this work; (b) SEM image of the prepared perovskite layer; (c) AFM height image of the prepared perovskite layer; (d) UV–vis absorption spectrum of the prepared perovskite layer.

260

X. Xie et al. / Chemical Physics Letters 687 (2017) 258–263

Fig. 2. J–V characteristics (a) and IPCE spectra (b) of PSCs prepared using spiro-OMeTAD layers processed from CB and DCB; forward and reverse J–V characteristics of PSCs prepared using spiro-OMeTAD layers processed from CB (c) and DCB (d).

Table 1 Properties of CB and DCB solvents and parameters (extracted from forward scanning J–V characteristics) of PSCs prepared using spiro-OMeTAD layers processed from CB and DCB. Solvent

Boiling point (°C)

Vapor pressure (kPa)

Evaporation rate (relative)

Voc (V)

Jsc (mA cm2)

FF (%)

PCE (%)

CB DCB

131 180.5

1.2 0.13

fast slow

1.03 ± 0.01 1.05 ± 0.01

19.6 ± 0.3 20.5 ± 0.3

71.9 ± 0.6 75.1 ± 0.5

14.5 ± 0.4 16.2 ± 0.3

3. Results and discussion The cross-section SEM images in Fig. S1(a)–(b) clearly highlight the FTO/TiO2/perovskite/spiro-OMeTAD/Au layer structure of PSCs based on spiro-OMeTAD processed from CB and DCB, respectively. The thicknesses of the perovskite and spiro-OMeTAD layers were estimated to be 400 and 220 nm, respectively. The quality of the perovskite layer was characterized using SEM, AFM and UV–vis measurements. The SEM and AFM height images in Fig. 1(b)–(c) reveals a full, pinhole-free, surface coverage of the perovskite film. The diameter of the perovskite particles is 200–500 nm, which is consistent with previous studies [3,8–10]. Fig. 1(d) shows a typical UV–vis absorption spectrum of the perovskite film in the 400– 800 nm region, highlighting the good light-harvesting properties of the perovskite layer [3–6]. The UV–vis absorption, SEM and AFM measurements demonstrate the good quality of the perovskite layer prepared in this study, which is an important requirement for fabricating high-performance PSCs. The forward scanning J–V characteristics of the PSCs prepared using spiro-OMeTAD layers processed from CB and DCB are shown in Fig. 2(a). The device parameters obtained from the J–V characteristics are listed in Table 1, which reports values averaged over 20 devices for each group. The PSCs prepared using CB for spiroOMeTAD processing exhibited an average PCE of 14.5%, with a Voc of 1.03 V, a Jsc of 19.6 mA cm2, and a FF of 71.9%, similar to the results obtained in previous studies of this structure [3–6]. When DCB was used for the deposition of spiro-OMeTAD, the

PCE increased to 16.2%, with a simultaneous enhancement in the Voc, Jsc, and FF values. Fig. 2(b) shows the IPCE results corresponding to the PSCs based on spiro-OMeTAD processed from CB and DCB. The Jsc parameters calculated by integrating the IPCE spectra of the CB- and DCB-based PSCs were 19.3 and 20.1 mA cm2, respectively. The difference between the Jsc values obtained from the IPCE and J–V measurements was only 1.5–2.0%, indicating the high accuracy and good reproducibility of our experiments [3– 8,11,12]. The highest PCE of 16.6% was obtained for the DCBbased group, along with a high Jsc of 20.7 mA cm2, a high FF of 75.8%, and a Voc of 1.06 V, which were comparable to the parameters reported for other high-performance PSCs with a similar structure [3–8]. Fig. 2(c) and (d) shows the hysteresis performance of PSCs based on spiro-OMeTAD layers processed from CB and DCB. The device parameters obtained under forward and reverse scans are summarized in Table 2. In both cases, the obtained PCEs were

Table 2 Parameters obtained for PSCs prepared using spiro-OMeTAD layers processed from CB and DCB, under forward and reverse scans. Solvent-scan

Voc (V)

Jsc (mA cm2)

FF (%)

PCE (%)

CB-forward CB-reverse DCB-forward DCB-reverse

1.03 1.04 1.05 1.05

19.6 19.6 20.5 20.5

71.9 74.6 75.1 75.5

14.5 15.2 16.2 16.3

X. Xie et al. / Chemical Physics Letters 687 (2017) 258–263

261

Fig. 3. AFM height images of spiro-OMeTAD layers cast from (a) CB and (b) DCB.

higher under reverse than forward scans, mainly due to the different FFs. When DCB was used for the preparation of spiro-OMeTAD, the hysteresis degree (difference between forward and reverse scans) was dramatically reduced from 4.8% to 0.6%, indicating a reduced number of defects in the corresponding PSCs [19–21]. Fig. S2 shows the steady-state PCE and current density as a function of time for PSCs based on spiro-OMeTAD processed from CB and DCB, respectively, indicating the stable outputs for both kinds of PSCs for a period of 100 s [12,23]. To determine the origin of the improved performance of devices based on DCB, we used AFM to inspect the surface morphology of spiro-OMeTAD layers processed from CB and DCB solvents, as shown in Fig. 3(a) and (b). The CB-processed spiro-OMeTAD layer had a rougher morphology, with a large number of particles on its surface [16,17]. The root-mean-square (RMS) roughness of the CB-processed spiro-OMeTAD layer was 8.7 nm. On the other hand, in the case of the spiro-OMeTAD layer processed from DCB, the surface showed a smoother morphology with only a few aggregates on the surface: as a result, the RMS roughness was considerably reduced to 3.5 nm. The parameters of the CB and DCB solvents, listed in Table 1, help explaining the observed differences in the surface morphologies. From Table 1, DCB has a higher boiling point and lower vapor pressure (corresponding to a lower evaporation rate) compared with CB [22]. As schematically illustrated in Fig. 4(a), during the spin coating process the spiro-OMeTAD layer changes from the liquid to the solid state. Because the evaporation rate of CB is higher, the solidification of spiro-OMeTAD occurs more rapidly in this case, resulting in a rougher surface morphology [11]. The particles observed on the CB- and DCB-processed spiro-OMeTAD surfaces (Fig. 4(a)) might consist of aggregates of spiro-OMeTAD with Li salt and/or 4-tert-butylpyridine [16,17]. The less volatile DCB solvent slowed down the solidification of spiro-OMeTAD, thus inducing a smoother surface morphology with Li salt and/or 4-tert-butylpyridine more uniformly dispersed in the spiro-OMeTAD layer [16,17]. As shown in Fig. S1, the smooth and uniform surface morphology of the spiro-OMeTAD layer further enhanced its contact with the Au electrode by reducing the pinholes at the interface, thus improving the carrier transport properties [3–7,16–21]. To verify the above hypothesis about the improved contact between the spiro-OMeTAD layer and the Au electrode, we measured the DC (direct current) conductivities of the spiro-OMeTAD layers processed from CB and DCB using an Au/spiro-OMeTAD/Au structure [23]. The conductivity (r) was calculated from the slope of the J–V plots using the relation J = rd–1V [14], where J is the current density, V is the applied voltage, and d is the thickness of the

Fig. 4. (a) Schematic illustration of the formation of spiro-OMeTAD layers processed from CB and DCB. (b) J–V characteristics of Au/spiro-OMeTAD/Au devices (shown in the inset) whose spiro-OMeTAD layers were processed from CB and DCB.

spiro-OMeTAD layers (200 nm). As shown in Fig. 4(b), conductivities of 0.025 and 0.032 mS cm–1 were estimated for spiroOMeTAD layers processed from CB and DCB, respectively [23]. The lower conductivity of the spiro-OMeTAD layer processed from CB probably originated from the high contact resistance between the rougher spiro-OMeTAD layer and the metal electrode. Moreover, the enhanced conductivity associated with the DCB-based layer can play an important role in the charge transport from perovskite to metal electrode in the corresponding PSCs, which is consistent with their improved Jsc and FF values [18–20,23]. On the other hand, a smooth and uniform spiro-OMeTAD layer results in a lower amount of defects in the PSCs. The number of defectinduced carrier traps at the interface was thus reduced, which explains well the reduced hysteresis [19–21]. We also performed

262

X. Xie et al. / Chemical Physics Letters 687 (2017) 258–263

Fig. 5. Stability test (normalized PCEs vs. time) of PSCs prepared using spiroOMeTAD layers processed from CB and DCB.

the PL test of bare perovskite, perovskite with CB processed spiroOMeTAD, and perovskite with DCB processed spiro-OMeTAD. As shown in Fig. S3, the PL intensity of perovskite with DCB processed spiro-OMeTAD is 23% lower than that of perovskite with CB processed case, indicating the improved quenching property at the interface between perovskite and spiro-OMeTAD by using DCB [23], which possibly induced by the improved contact of perovskite/ spiro-OMeTAD. In addition, we investigated the stability of PSCs based on the CB- and DCB-processed spiro-OMeTAD layers by storing the unencapsulated PSCs in air under 40% humidity [6,24]. Fig. 5 shows that the PSCs based on the CB- and DCB-processed spiroOMeTAD layers still exhibited PCEs of 11.5% and 13.8% (corresponding to degradations of 21% and 15%), respectively, after 7 d. This indicates that a smoother and uniform spiro-OMeTAD layer is advantageous for the long-term operation of PSCs, possibly because of the good spiro-OMeTAD coverage on the perovskite layer, which efficiently protects the perovskite layer from atmospheric moisture. Fig. 6(a) shows the EIS results obtained for the PSCs prepared using CB- and DCB-processed spiro-OMeTAD layers. The smaller semicircle at the high frequency presents the contact resistance (Rco) of perovskite/TiO2 or perovskite/spiro-OMeTAD, while the larger semicircle at the low frequency is assigned to the interface and/ or bulk recombination (Rrec) [10,25]. The parameters by fitting the equivalent circuit in Fig. 6(a) are summarized in Table 3. The Rco of the PSCs based on CB and DCB processed spiro-OMeTAD layers are 53.5 and 48.3 X, respectively. Since both kinds of the PSCs are fabricated on the same TiO2 layers, the decreased Rco indicates the improved hole extraction efficiency at the interface of perovskite/ spiro-OMeTAD by using DCB [10,25], which is in good agreement with the DC conductivity and AFM analysis. Moreover, the Rrec of the PSCs using DCB processed spiro-OMeTAD is 1709.3 X, much higher than that (1533.6 X) of PSCs based on CB, indicating a reduced recombination at the interfaces of perovskite/spiroOMeTAD and/or spiro-OMeTAD/Au electrode contacts due to the smoother spiro-OMeTAD layer [10,25]. Fig. 6(b) shows the dark J–V characteristics of PSCs incorporating spiro-OMeTAD layers processed from the different solvents. The other device parameters could be obtained by fitting the curves in Fig. 6. In the equivalent circuit model, the J–V characteristics are described by the following equation [11,26–28]:

lnðJÞ ¼ lnðJ 0 Þ þ

  1 q V n kB T

ð1Þ

Fig. 6. (a) EIS analysis and (b) dark J–V characteristics of PSCs based on spiroOMeTAD layers processed from CB and DCB.

Table 3 Fitted parameters for PSCs based on spiro-OMeTAD layers processed from CB and DCB solvents. Solvent

Rs (X)

Rco (X)

Rrec (X)

CB DCB

30.6 29.7

53.5 48.3

1533.6 1709.3

where J0 is the reverse saturation current density, n is the ideality factor, q is the elementary charge, kB is the Boltzmann constant, T is the absolute temperature, and A is the area of the solar cell. The results of the fitting procedure show that J0 and n decreased from 1.35  10–6 mA cm2 and 2.33 to 7.62  10–7 mA cm2 and 2.30, respectively, when DCB was used instead of CB. Since J0 and n reflect the leakage current due to carrier and charge recombination in the space-charge region [11,26–28], the decreased J0 and n values indicate a suppressed charge recombination due to the smoother spiro-OMeTAD layers, in good agreement with the EIS analysis. Moreover, the suppression of charge recombination can also explain the enhanced Jsc and FF values of PSCs [11,26–28]. Furthermore, under open-circuit conditions, Eq. (1) can be simplified into the following Shockley equation [11,26–28]:

V oc 

nkB T J lnð sc Þ q J0

ð2Þ

Therefore, the lower J0 and n values, together with the improved Jsc, can explain the enhanced Voc observed for PSCs based on DCB.

X. Xie et al. / Chemical Physics Letters 687 (2017) 258–263

4. Conclusions In conclusion, we investigated the effect of the spiro-OMeTAD morphology on the performance of PSCs prepared using CB or DCB for the deposition of the spiro-OMeTAD layer. We found that DCB led to a smoother spiro-OMeTAD layer compared with CB. This is due to the low volatility of DCB, which suppresses solvent evaporation during the formation of the spiro-OMeTAD layer (solidification during spin coating). The improved morphology of the spiro-OMeTAD layer improves its contact with the perovskite layer and the metal electrode. The use of DCB for processing spiroOMeTAD led to an increase from 14.5% to 16.2% in the PCE of the PSCs, accompanied by a simultaneous enhancement in the Voc, Jsc, and FF parameters. Furthermore, the hysteresis was dramatically reduced from 4.8% to 0.6% and the stability in air was also improved. EIS and dark current analyses indicated a suppression in charge recombination due to the improved morphology obtained using DCB. A high PCE of 16.6% was achieved for the best PSC based on the DCB-processed spiro-OMeTAD layer. Our results suggest that the use of DCB for spiro-OMeTAD processing represents a simple and effective way for fabricating highperformance and air-stable PSCs with low hysteresis. Acknowledgement This work was supported by the Department of Science & Technology of Jilin Province, China (Developmental Project of Science and Technology of Jilin Province, Grant No. 20140204086GX). Appendix A. Supplementary material Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/j.cplett.2017.09. 031. References [1] A. Kojima, K. Teshima, Y. Shirai, T. Miyasaka, J. Am. Chem. Soc. 131 (2009) 6050–6051.

263

[2] National Renewable Energy Laboratory chart, http://www.nrel.gov/ncpv/ images/efficiency_chart.jpg, accessed 03/13/2016. [3] M.M. Lee, J. Teuscher, T. Miyasaka, T.N. Murakami, H.J. Snaith, Science 338 (2012) 643–647. [4] M. Liu, M.B. Johnston, H.J. Snaith, Nature 501 (2013) 395–398. [5] H.-S. Kim, C.-R. Lee, J.-H. Im, K.-B. Lee, T. Moehl, A. Marchioro, S.-J. Moon, R. Humphry-Baker, J.-H. Yum, J.E. Moser, M. Grätzel, N.-G. Park, Sci. Rep. 2 (2012) 591–597. [6] H. Zhou, Q. Chen, G. Li, S. Luo, T. Song, H.-S. Duan, Z. Hong, J. You, Y. Liu, Y. Yang, Science 345 (2014) 542–546. [7] H. Kim, K.-G. Lim, T.-W. Lee, Energy Environ. Sci. 9 (2016) 12–30. [8] Z. Xiao, Q. Dong, C. Bi, Y. Shao, Y. Yuan, J. Huang, Adv. Mater. 26 (2014) 6503– 6509. [9] Y. Shao, Y. Yuan, J. Huang, Nat. Energy 1 (2016) 15001. [10] Y. Yang, K. Ri, A. Mei, L. Liu, M. Hu, T. Liu, X. Li, H. Han, J. Mater. Chem. A 3 (2015) 9103–9107. [11] Z. Liu, E.-C. Lee, Org. Electron. 24 (2015) 101–105. [12] K. Chen, Q. Hu, T. Liu, L. Zhao, D. Luo, J. Wu, Y. Zhang, W. Zhang, F. Liu, T.P. Russell, R. Zhu, Q. Gong, Adv. Mater. 28 (2016) 10718–10724. [13] M. Salado, M. Oliva-Ramirez, S. Kazim, A.R. González-Elipe, S. Ahmad, Nano Energy 35 (2017) 215–222. [14] J. Choi, S. Song, M.T. Hörantner, H.J. Snaith, T. Park, ACS Nano 10 (2016) 6029– 6036. [15] X. Wang, Z. Li, W. Xu, S.A. Kulkarni, S.K. Batabyal, S. Zhang, A. Cao, L.H. Wong, Nano Energy 11 (2015) 728–735. [16] E.J. Juarez-Perez, M.R. Leyden, S. Wang, L.K. Ono, Z. Hawash, Y. Qi, Chem. Mater. 28 (2016) 5702–5709. [17] M.-C. Jung, S.R. Raga, L.K. Ono, Y. Qi, Sci. Rep. 5 (2015) 9863. [18] T. Bu, L. Wu, X. Liu, X. Yang, P. Zhou, X. Yu, T. Qin, J. Shi, S. Wang, S. Li, Z. Ku, Y. Peng, F. Huang, Q. Meng, Y.-B. Cheng, J. Zhong, Adv. Energy Mater. (2017) 1700576. [19] S. Ryu, J. Seo, S.S. Shin, Y.C. Kim, N.J. Jeon, J.H. Noh, S.I. Seok, J. Mater. Chem. A 3 (2015) 3271–3275. [20] Y. Shao, Z. Xiao, C. Bi, Y. Yuan, J. Huang, Nat. Commun. 5 (2014) 5784. [21] Z. Xiao, C. Bi, Y. Shao, Q. Dong, Q. Wang, Y. Yuan, C. Wang, Y. Gao, J. Huang, Energy Environ. Sci. 7 (2014) 2619–2623. [22] I.M. Smallwood, Handbook of Organic Solvent Properties, Halsted Press, New York, 1996. [23] J.H. Heo, H.J. Han, D. Kim, T.K. Ahn, S.H. Im, Energy Environ. Sci. 8 (2015) 1602– 1608. [24] L. Chen, X. Xie, Z. Liu, E.-C. Lee, J. Mater. Chem. A 5 (2017) 6974–6980. [25] M. Hu, L. Liu, A. Mei, Y. Yang, T. Liu, H. Han, J. Mater. Chem. A 2 (2014) 17115– 17121. [26] H.-W. Kang, J.-W. Lee, D.-Y. Son, N.-G. Park, RSC Adv. 5 (2015) 47334–47340. [27] S.-H. Oh, S.-I. Na, J. Jo, B. Lim, D. Vak, D.-Y. Kim, Adv. Funct. Mater. 20 (2010) 1977–1983. [28] X. Xie, G. Liu, C. Xu, S. Li, Z. Liu, E.-C. Lee, Org. Electron. 44 (2017) 120–125.