Perovskite Solar Cells
www.afm-journal.de
Solvent-Assisted Low-Temperature
Crystallization of SnO 2 Electron-Transfer Layer for High-Efficiency Planar Perovskite Solar Cells
Cong Chen, Yue Jiang,* Jiali Guo, Xiayan Wu, Wenhui Zhang, Sujuan Wu, Xingsen Gao, Xiaowen Hu, Qianming Wang, Guofu Zhou, Yiwang Chen, Jun- Ming Liu, Krzysztof Kempa, and Jinwei Gao*
C. Chen, Dr. Y. Jiang, J. Guo, X. Wu, W. Zhang, Prof. S. Wu, Prof.
X. Gao, Prof. J.-M. Liu, Prof. K. Kempa, Prof. J. Gao Institute for Advanced Materials and Guangdong Provincial Key Laboratory of Quantum Engineering and Quantum Materials Academy of Advanced Optoelectronics South China Normal University
Guangzhou 510006, China
E-mail: [email protected]; [email protected] Dr. X. Hu, Prof. G. Zhou
Guangdong Provincial Key Laboratory of Optical Information Materials and Technology & Institute of Electronic Paper Displays South China Academy of Advanced
Optoelectronics South China Normal University Guangzhou 510006, China
Prof. Q. Wang
School of Chemistry and Environment South China Normal University Guangzhou 510006, China Prof. Y. Chen
Institute of Polymers and Energy Chemistry College of Chemistry Nanchang University
Nanchang 330031, China Prof. J.-M. Liu
Laboratory of Solid State Microstructures Nanjing University Nanjing 210093, China
Prof. K. Kempa Department of Physics Boston College
Chestnut Hill, MA 02467, USA
The ORCID identification number(s) for the author(s) of this article can be found under https://doi.org/10.1002/adfm.201900557.
DOI: 10.1002/adfm.201900557
1. Introduction
Lead halide perovskite solar cells (PSCs) have attracted great attention due to simplicity, and low cost of fabrication, with rapidly improving power conver-sion efficiency, recently reported as high as 23.3%.[1] This rapid progress has been mainly due to the extensive studies of perovskite films,[2] electron transfer layers[3] (ETL), as well as the hole trans- port layers (HTL).[4] The main goal was to obtain better crystallinity of the perov-skite films,[5] higher charge transport effi- ciencies,[6] better film morphologies, and enhanced interfacial quality,[7] The n-i-p type PSCs, with the configuration FTO/c- TiO2/m-TiO2/CH3NH3PbI3 (MAPbI3)/spiro-OMeTAD/Ag, have been widely studied, mostly because of the maturity of its fab- rication procedure.[8] In those devices TiO2 is commonly used as an ETL, which can provide good energy level matching and a fair electron mobility (10−5 cm2 V−1 s−1). However, its high- temperature sintering process (450 C) is not compatible with low-cost flexible substrates and roll-to-roll processing. Mean- while, there have been efforts to develop simpler devices even without the mesoporous layer, which however suffer from lower efficiency.[9]
SnO2 is an excellent alternative to TiO2, which has a wide optical bandgap (3.6–4.0 eV), and a much higher electron mobility (up to 240 cm2 V −1 s−1).[10] In addition, it can be pro-cessed at low-temperature, and has a good chemical and optical stability.
Low-temperature atomic layer deposition (ALD) at 120 C has been used to deposit a layer of SnO2 as ETL, with the PCE above 18%, whereas this film is generally amorphous and limits its electron mobility.[11]
Despite, plasma-assisted ALD enhanced the crystallinity of SnO2, the necessary fabrication temperature was as high as 200
C.[12] In another attempt, sol-gels of SnCl2,·2H2O, or SnCl4·5H2O was spin-coated, followed by ≈180 C anneal.[13]
This process produced a high-quality conductive film, but the relatively high annealing temperature made this process incompatible with most of plastic substrates. Moreover, the influence of thermal annealing temperature on the flexible sub- strates has been systematically investigated[14] demonstrating that the processing temperature of 150 C is still incompatible
A high-quality polycrystalline SnO2 electron-transfer layer is synthesized through an in situ, low-temperature, and unique butanol–water solvent- assisted process. By choosing a mixture of butanol and water as a solvent, the crystallinity is enhanced and the crystallization temperature is lowered to 130 C, making the process fully compatible with flexible plastic substrates. The best solar cells fabricated using these layers achieve an efficiency of 20.52% (average 19.02%) which is among the best in the class of planar n–i– p-type perovskite (MAPbI3) solar cells. The strongly reduced crystallization temperature of the materials allows their use on a flexible substrate, with a resulting device efficiency of 18%.
achieving an efficiency as high as 20.52% (averaged 19.02%) of the resulting n-i-p planar PSCs (based on the CH3NH3PbI3). This annealing temperature is acceptable for a whole class of inexpensive plastic, flexible substrates.
2. Results and Discussion
2.1. Crystallization and Morphology of SnO2 ETLs
To obtain highly crystallized SnO2 ETLs, the solvent used is of great importance for the hydrolysis process due to its direct and fundamental influence on the quality of SnO2 NCs and the following thermal annealing process. Hence, four types of alcohol are selected with the consideration of their increasing boiling point (ethanol (78 C) isopropanol (82 C) isobu-tanol (107 C)
butanol (117 C)) and a small amount of water was added in order to enhance the hydrolysis efficiency. We designate SnO2 ETLs synthesized from ethanol, isopro-panol, isobutanol, butanol, and mixture of butanol-5 vol% water as E-SnO2, iP-SnO2, iB- SnO2, B-SnO2, and WB-SnO2 correspondingly.
SnO2 NCs were synthesized with a hydrolysis reaction in solvent through refluxing at 110 C.[18] Then the SnO2 ETLs were obtained by spin-coating the as-obtained SnO2 NCs solu-tion on FTO substrate, followed by 1 h of thermal annealing at 130 C.[18]
This thermal annealing temperature was chosen based on our preliminary evaluations described in the Sup-porting Information.
Figure 1a–e show high-resolution transmittance electron microscopy (HRTEM) images of the corresponding films. Insets show corresponding electron dif-fraction patterns. For single solvent, the best crystallinity has the film reacted in butanol (Figure 1d).[19] The d-spacing was measured to be 0.33 nm (Figure 1c,d), corresponding to the interplanar spacing of (110) plane of the rutile phase of SnO2. The NCs diameter is about 3–5 nm for iB-SnO2 and B-SnO2 NCs. Figure 1f–i show X-ray diffraction (XRD) patterns with diffraction peaks assigned to the (110), (101), (200), (211), (220), (310), and (301) planes of SnO2,[19] with increased intensity in the range of B-SnO2 iB- SnO2 iP-SnO2 E-SnO2, confirming the higher crystallinity of B-SnO2 ETL as indicated in TEM images (in Figure 1).
X-ray photoelectron spectroscopy (XPS) measurement con- firms the composition of the obtained SnO2 ETLs (Figure S1,
Figure 1. HRTEM images and XRD patterns of SnO2 nanocrystals.
a) and f) E-SnO2. b) and g) iP-SnO2. c) and h) iB-SnO2. d) and i) B-SnO2, and e) and j) WB-SnO2. Insets show corresponding electron diffractions.
Scale bars are 5 nm.
Supporting Information). The full XPS spectrum survey given in Figure S1a, Supporting Information, shows the presence of O and Sn. The peaks at binding energies of 487.20, 487.23,
487.25 eV belongs to Sn 3d5/2 and of 495.60, 495.62, 495.62 eV are contributed by Sn 3d3/2, respectively, corresponding to E-SnO2, B-SnO2, and WB-SnO2 (Figure S1b, Supporting Infor- mation). The peaks of O 1s are located at binding energy of 531.15, 531.21, 531.20 eV, attributed to the O2− state in SnO2 (Figure S1c, Supporting Information).[20] Moreover, there is almost no Cl residual observed in WB-SnO2 film, and Cl con-tent in B-SnO2 and E-SnO2 film is gradually increased as illus-trated by the higher peaks at binding energy from 198 to 200 eV (Figure S1d, Supporting Information).
The morphologies of the different SnO2 ETLs placed on glass/FTO substrates were characterized with scanning electron microscopy (SEM) images, shown in Figure 2a–e. As shown in Figure 2a, E-SnO2 film has some of the blur spots and pinholes; iP- SnO2 has some pinholes (Figure 2b); iB- SnO2 also presents some blur (Figure 2c). Those results mean that the crystallinity of E- SnO2, iP-SnO2, and iB-SnO2 is inferior as confirmed in XRD. By contrast, B-SnO2 ETL, shown in Figure 2d is the best, with a uniform, continuous, and compact morphology, as well as high crystallinity. Thereby, the remarkably improved mor-phology quality of B-SnO2 ETLs should be ascribed to the fact that the higher boiling point of butanol allows a slow solvent evaporation process, providing sufficient time for SnO2 NC re-crystallization.
It is well known that water is one of the key reactants during SnCl2 hydrolysis process, and usually only the moisture in the air is utilized. Herein, instead of using dry solvent, we added
Figure 2. Top-view SEM images of ETLs. a) E-SnO2, b) iP-SnO2, c) iB-SnO2, d) B-SnO2, e) WB-SnO2, and f) FTO. Scale bars are 200 nm.
5% v/v of deionized (DI) water into the butanol. A series of water quantities, 5%, 10%, and 15% (water and butanol are no longer mutual soluble above the ration of 15%), has been investigated as shown in Figure S2, Supporting Information, demonstrating the limited influence of water quantity on the PCE of the PSCs. The volume percentage of 5% was chosen on account that this quantity is enough for the reaction, and thus the basicity of the final mixed solvent would be barely influ-enced. With extra water dissolved in butanol, the hydrolysis reac-tion rate was increased with reaction time shortened from 13 to 4 h. The crystallinity of SnO2 was further improved as illustrated by the TEM in Figure 1e and XRD in Figure 1j. Moreover, the added water improved the quality of SnO2 ETL (WB-SnO2), to a uniform, pinhole-free morphology (Figure 2e). As shown in Figure 2, SnO2 with butanol solvent shows the best morphology. Therefore, in the following sections, E-SnO2, B-SnO2, and WB-SnO2 were selected as the prototypes to discuss the influence of solvent and the extra water on the optoelectronic properties. The morphology of bare FTO as reference was shown in Figure 2f.
2.2. Charge Transfer Properties and Optical Transmittance
Electronic structures of the valence bands were measured using UV photoelectron spectrometer (UPS) (Figure 3a). The Fermi Levels (EF) of −4.37, −4.25, and −4.25 eV for E-SnO2, B-SnO2, and WB-SnO2, respectively were estimated from the intercept of the higher binding energy with incident photon energy, while the corresponding valence band maximum (VBM) are −8.31, −8.21, and −8.25 eV calculated from the intercept of the lower binding energy with EF.[21] Optical band gap (Eg) evaluated by UV–vis absorption spectroscopy (Figure S3a, Supporting Information) are 4.39, 4.39, 4.41 eV. Hence it can be calculated that the conduction band minimum (CBM) are located at −3.92, −3.82, −3.84 eV for E- SnO2, B-SnO2, and WB-SnO2, respectively (Figure S3a, Supporting Information). Thus, it is obvious that B-SnO2 and WB- SnO2 are not only well matched to the perovskite layer but also expected to favor a higher VOC (Figure S3b, Supporting Information).[22] The electron mobility is another key factor for ETLs, which was
assessed by the space-charge limited current (SCLC), achieving 9.45 10−5, 2.99 10−4, 6.02 10−4 cm2V−1 s−1 for E-SnO2, B- SnO2, WB-SnO2, respectively (Figure 3b), which are con-sistent with results from literature.[23] The optical transmittance of different SnO2 is also measured (Figure S4, Supporting Information). There is no different in transmittance for all of those SnO2, which is consistent with optical absorption (shown in Figure S3a, Supporting Information).
The charge transfer efficiency between perovskite absorber and the SnO2-based ETL was investigated by steady-state photoluminescence spectroscopy (PL) and time-resolved photoluminescence (TRPL) on the films of perovskite, E-SnO2/ perovskite, B-SnO2/perovskite, and WB-SnO2/perovskite deposited on quartz (Figure 3c,d). The emission peak at 770 nm originates from the MAPbI3, which is significantly quenched with WB-SnO2 ETL. It demonstrates the fast electron extrac-tion and transportation from perovskite layer, agreed well with its higher electron mobility (Figure 3b). TRPL has further con-firmed this higher electron transfer efficiency with the reduced PL decay time (τ1 and τ2) shown in Table S1, Supporting Information. τ1 of 10.22 ns for ETL-free sample was gradu-ally reduced to 6.55, 6.17, and 4.72 ns when applying E-SnO2, B-SnO2, and WB-SnO2 ETLs, respectively. Moreover, the same trend was observed in respect of τ2 (77.44, 38.58, 31.56, and 17.47 ns), which is because of the enhanced charge transfer efficiency at SnO2/perovskite interfaces. Compared with WB-SnO2, the less efficient charge transport process between E-SnO2 or B-SnO2 ETLs with perovskite layer is caused by the interfacial traps introduced by the less well-formed morphology and the poor crystallinity (Figures 1 and 2). Specifically, the charge transfer rate (Kct)[24] was studied using the difference of reciprocal global lifetime for perovskite with different ETLs. The faster charge transfer rate of WB-SnO2 (0.074 ns−1) than B-SnO2 (0.021 ns−1) and E-SnO2 (0.014 ns−1) agrees well with the PL decay time.
Surface potential measured by scanning Kelvin probe micros- copy (SKPM) in Figure 4a–c reveals a higher and homogeneous surface potential profile of the WB-SnO2 film (686.15 mV) than those of E-SnO2 (419.77 mV) and B-SnO2 (575.25 mV). This fact means a lower work function and shallower Femi level
Figure 3. Characterization and charge transfer properties of the ETLs. a) UPS of SnO2 deposited on quartz. b) SCLC of FTO/ETLs/Ag. c) Steady-state PL and d) TRPL of FTO/ETL/MAPbI3.
for WB-SnO2,[25] which agrees well with the EF from UPS. The conductive atomic force microscopy (CFM) was also performed on the SnO2 films. As shown in Figure 4d–f, the average
currents are counted to be 732 pA, 6 nA, and 12 nA for E-SnO2, B- SnO2, and WB-SnO2 ETLs, respectively, which shows that the conductivity of WB-SnO2 is the best.[26]
Figure 4. SKPM and CFM images of the ETLs. a–c) SKPM and d–f) CFM images of a,d) E-SnO2, b,e) B-SnO2, and c,f) WB- SnO2.
Figure 5. Performance of SnO2 based PSCs. a) Cross-sectional SEM image of the completed devices, with the structure Glass/FTO/SnO2/MAPbI3/ Spiro-OMeTAD/Ag. b) Top-view SEM images of perovskite deposited on WB-SnO2, c) J–V curves with the inset showing parameter values. d) EQE (left panel) and current density (right panel) of different SnO2 based PSCs.
It is reported that the mixture of crystalline and amor-phous SnO2 leads to oxygen-vacancy-related defects, which could capture electrons and cause recombination at the SnO2/ perovskite interface.[16] Thus, the observed better electron mobility, conductivity, and transfer efficiency of WB-SnO2 is reasonably attributed to its continuous, crystalline, and uniform morphology.
2.3. Perovskite Solar Cells
A series of n-i-p type planar PSCs based on various SnO2 ETLs, with configuration of FTO/SnO2/MAPbI3/spiro-OMeTAD/ Ag was fabricated, as shown in cross-sectional SEM image in Figure 5a. To exclude possible processing variations, the devices were fabricated in the same batch, with the exactly identical procedures and parameters. The thickness of SnO2, perovskite, spiro- OMeTAD, and Ag electrode was about 40, 400, 200, and 70 nm, respectively. Meanwhile, the thickness of E-SnO2, B-SnO2, and WB-SnO2 were controlled to be nearly equal, ≈40 nm. The morphologies of perovskite films deposited on top of the SnO2 are shown in SEM images in Figure 5b and Figure S5, Supporting Information. It is clear, B-SnO2 and WB-SnO2 based perovskite layers are pinhole-free, while those cells based on E-SnO2 are full of pinholes and vacancies.
The J–V curves with the parameters of the best devices are shown in Figure 5c. The E-SnO2 based device had a power conversion efficiency (PCE) of 14.26% with open-circuit voltage (VOC) of 1.08 V, short-circuit current density (JSC) of 21.30 mAcm−2, and fill factor (FF) of 61.89%. This is slightly improved by the application of iP-SnO2 and iB-SnO2 ETLs
(Figure S6, Supporting Information). However, a significant improvement is achieved when B-SnO2 is adopted, showing PCE of 19.70% (VOC of 1.09 V, JSC of 22.85 mAcm−2, FF of 79.16%). As expected, the record high a PCE of 20.52% (VOC as high as 1.10 V, JSC of 22.98 mA cm−2, and FF of 81.27%) was achieved with the WB-SnO2 ETL. The external quantum
efficiency (EQE) spectra shown in Figure 5d are consistent with the above results.
Figure 6a shows the steady-state PCE and current density of our PSCs under the constant bias voltages, at the maximum power output points (Vmp, 0.76, 0.89, 0.93 V) in J–V curves. The stabilized PCEs are 14.93%, 18.40%, and 19.62% with the current density of 19.52, 20.75, and 21.10 mA.cm−2 based on E-SnO2, B- SnO2, and WB-SnO2, respectively, which agrees well with the highest PCEs data. Their response time was measured to be 15.96, 10.92, and 7.80 s, indicative of the reduced defect of WB-SnO2 ETL. The J–V curves of PSCs based on WB-SnO2 ETL from forward and reverse scan are shown in Figure S7, Supporting Information. Around 1% PCE difference could be observed, which is probably related to the interfacial charge accumulations, ion migration in perovskite, ferroelectric polarization, etc.[27]
Whereas, the steady state PCE presenting short response time is merely ascribed to the relieved charge accumulation at the ETL/perovskite interface. Additionally, the JSC estimated from J–V curve is in agreement with JSC calcu-lated from EQE, further supporting the reliable performances of our Perovskite solar cells.[28]
To study the reproducibility of the devices, 24 PSCs for each ETL were fabricated and characterized. The param-eter distribution histograms of cell performance are shown in Figure 6b, Figure S8 and Tables S2–S4, Supporting
