High-Quality Perovskite CH3NH3PbI3 Thin Films for Solar Cells Prepared by Single-Source Thermal Evaporation Combined with Solvent Treatment

In this work, solvent annealing process for CH3NH3PbI3 thin film prepared by single source evaporation was reported. Characterized by the scanning electron microscope (SEM), X-ray diffractometer (XRD), energy dispersive spectroscope (EDS), ultraviolet-visible (UV) spectrophotometer, and the photoluminescence (PL) spectrometer, our method ensured higher quality film with crystallinity, composition, well-defined grain structure, and reproducibility. The optimized solar cell device based on the structure of ITO/PEDOT:PSS/CH3NH3PbI3/PCBM/Ag achieved better performance in power conversion efficiency from 2.64% to 9.92%, providing an effective method to optimize the quality of perovskite film for solar cell application.


Introduction
In recent years, owing to their low exciton binding energy, high carrier mobility, long carrier diffusion length, wide absorption spectrum, high optical absorption rate, and controllable composition, organic and inorganic hybrid metal halide perovskite materials have attracted much attention and quickly became one of the promising materials for solar cell application [1][2][3]. Their molecular formula could be expressed as ABX 3 (A represents organic cation CH 3 NH 3 + , HC(NH 2 ) 2 + , CH 3 CH 2 NH 3 + , or C(NH 2 ) 3 + ; B represents metal cation Pb 2+ or Sn 2 ; X represents halogen anion I − , Cl − , or Br − ). It is unbelievable that the power conversion efficiency (PCE) of MAPbI 3 solar cell has increased dramatically among all kinds of solar cells. Within a short time, from 2009 to 2016, the PCE of perovskite solar cells based on organic and inorganic hybrid metal halide perovskite (CH 3 NH 3 PbI 3 (MAPbI 3 )) as absorber increases rapidly from 3.8% to 22.1%. Moreover, the latest research proved that the PCE of MAPbI 3 solar cells could rise to 31% [4], which indicates that the development of MAPbI 3 solar cells is not only limitless, but also has great potential for further exploration in solar cell application. As we all know, the thickness [5,6], size of crystal grains [7], elements stoichiometric ratio [8,9], uniformity, and coverage [10,11] of light absorption layer have a great influence on the performance of MAPbI 3 solar cells, which directly affects the efficiency of perovskite solar cells. As for MAPbI 3 absorber, the surface morphology of the thin film is much sensitive to the preparation processes. Recently, many fabrication methods have been applied for the preparation of the perovskite thin films, such as one-step solution, two-step solution, vacuum dual-source co-evaporation, vapor-assisted deposition and vacuum single-source thermal evaporation. Compared to solution methods, vacuum evaporation is not involved with solvent evaporation so that it could easily regulate and control preparation process and obtain the perovskite films with high homogeneity and uniformity. In this work, we adopted the vacuum single-source thermal evaporation, proposed by Fan et al. [12][13][14][15], to fabricate MAPbI 3 films. Although the thin film was continuous with complete coverage, its crystal grains were small. Researchers found that perovskite films with micron-sized crystal grains is favorable for the carriers to separate and transport in the thin film [1,7,[16][17][18][19][20] and solvent annealing is an effective method to promote the growth of the crystal grains [21,22]. For the sake of promoting the growth of grains, solvent annealing for MAPbI 3 films with 1,4-Butyrolactone under atmospheric environment is an effective method [21,22]. In the air environment, the as-prepared perovskite films were annealed under solvent 1,4-Butyrolactone vapor and the solvent-annealed perovskite films with micron-sized grains were obtained. The effect of solvent-annealing process on the microstructure, composition, crystallinity, and optical properties of MAPbI 3 thin films were investigated. Based on the solvent-annealed thin films, the perovskite solar cell achieved an optimized PCE from 2.64% to 9.92%.

CH 3 NH 3 I synthesis:
We optimized the preparation process of CH 3 NH 3 I according to the references [23]. First, 28 mL methylamine (CH 3 NH 2 ) (40 wt.% in water, Sigma Aldrich, St. Louis, MI, USA) reacted with 30 mL hydroiodic acid (HI) (57 wt.% in water, Sigma Aldrich) in the 250 mL round flask at 0 • C with constant magnetic stirring for 2 h. Then, we removed the reacted solution to the 500 mL beaker fixed on the water bath heated at 90 • C to attain white crystal power. Subsequently, the powder was purified using anhydrous ethanol and ether for three or four times. Last, the CH 3 NH 3 I powder was collected in the brown bottle and saved in the glove box before use after it was dried in the vacuum oven at 60 • C or 24 h. Preparation of CH 3 NH 3 PbI 3 powder: First, we weighed 9.22 g PbI 2 powder and 3.14 g CH 3 NH 3 I powder, respectively, using the electronic balance and put them into the 20 mL γ-butyrolactone (≥99%, Sigma-Aldrich). Then, we removed the solution to the petri dish after ultrasonication for 30 min and heat it on the plate at 160 • C for 4 h. Finally, the CH 3 NH 3 PbI 3 powder was collected in the brown box and stored in the glove box before use.

Film Preparation
First, the blank glass substrates were sequentially cleaned in the ultrasonic machine with deionized water, acetone and ethanol. Then, the glass substrates were dried by the nitrogen gun and stored in the drying oven for use. Later, we placed them into the substrate holder, close its baffle, and a certain amount of CH 3 NH 3 PbI 3 powder was put into the evaporating boat. The deposition process of the CH 3 NH 3 PbI 3 films is as follows. We turned on the heating power when the vacuum pressure of the chamber was down to 1.5 × 10 −3 Pa. Then, we opened the baffle, smoothly increased the heating power to obtain the desired evaporating rate, and held it there for 15 min. The vertical distance between substrate holder and the evaporating boat was around 12 cm. After the perovskite films were done, they were thermally annealed at 150 • C for 20 min in a container of 500 mL under atmospheric environment. Meanwhile, 20 uL solvent γ-butyrolactone was evaporated in the closed container.

Device Fabrication
ITO (10 Ohm/sq, YINKOU OPVTECH, Yingkou, China) glass substrates were cleaned in the ultrasonic cleaning machine sequentially with the mixed solution of Decon 90, deionized water, and ethanol for 30 min, respectively. Then, the ITO substrates were dried by the nitrogen gun and stored in the drying oven for use. Prior to spin-coating the hole transport layer of PEDOT:PSS (CLEVIOS PVP AI4083, 70 nm), the ITO substrates were treated in the UV ozone cleaning system for 30 min. Then, the PEDOT:PSS layer was spin-coated on the substrates at 3000 rpm for 30 s. After that, the PEDOT:PSS-coated substrates were thermally annealed on a hot plate under the temperature of 180 • C for 20 min. The perovskite films were prepared by single-source vacuum thermal evaporation and the detailed information is in the Film Preparation. The electron transport layer (60 nm) of PCBM was spin-coated on the perovskite film at 3000 rpm for 30 s. The PCBM solution was prepared by dissolving 20 mg PCBM (>99%, Xi'an Polymer Light Technology Corp., Xi'an, China) powder in 1 mL chlorobenzene. Finally, a 150 nm thickness silver layer was deposited on the top of the device by thermal evaporation under the pressure of 1.5 × 10 −3 Pa.

Characterization
The surface morphology of perovskite films was measured by high-resolution scanning electron microscope ( Figure 1 shows the XRD pattern of the CH 3 NH 3 PbI 3 powder and its corresponding data of tetragonal phase. The XRD result of CH 3 NH 3 PbI 3 powder was in well agreement with the CH 3 NH 3 PbI 3 JCPDS data, indicating that CH 3 NH 3 PbI 3 powder with tetragonal phase was obtained by our method in the Experimental details. Figure 2 shows the surface and cross-section morphology of as-prepared and solvent-annealed CH 3 NH 3 PbI 3 films. All the CH 3 NH 3 PbI 3 thin films exhibited full surface coverage. However, the average grain size was mostly below 500 nm for as-prepared thin films while the solvent-annealed was up to 1 µm. The perovskite films were thermally annealed in the solvent vapor atmosphere in a sealed container. With the suitable usage of solvent evaporated, the solvent vapor could go into the grain boundaries and partially dissolve the perovskite film. Meanwhile, the perovskite films could recrystallize as the solvent was re-evaporated under the high annealing temperature. In addition, the solvent-annealed film showed a smoother surface with bigger crystal grains compared to as-prepared, which could lead to fewer crystal boundaries and accelerate the charge extraction process. From the cross-section images of CH 3 NH 3 PbI 3 films, it can be found that the larger grains were evenly distributed with the size over film thickness which was beneficial to significantly decrease charge recombination loss between boundaries. Figure 3 shows the X-ray diffraction (XRD) pattern and its corresponding FWHM and average crystal size of as-prepared and solvent-annealed CH 3 NH 3 PbI 3 films. All the samples exhibited intense diffraction peaks at 14.02 • , 28.32 • , and 31.76 • , respectively assigned to the (110), (220), and (310) planes of tetragonal perovskite structure [24], which indicated that the desired perovskite film was formed. Obviously, the main peak intensity of the solvent-annealed CH 3 NH 3 PbI 3 film was stronger and sharper than that the as-prepared, which was consistent with the trend of FWHM and shown in Figure 3b. The average crystalline size of (110), (220), and (310) planes in the as-deposited and solvent-annealed film was estimated by resolving the characteristic peaks using Scherrer's formula where K is Scherrer constant, D is the average grain size, B is the half-width of the measured diffraction peak, θ is the diffraction angle, and γ is the X-ray wavelength (0.154056 nm).   Figure 3 shows the X-ray diffraction (XRD) pattern and its corresponding FWHM and average crystal size of as-prepared and solvent-annealed CH3NH3PbI3 films. All the samples exhibited intense diffraction peaks at 14.02°, 28.32°, and 31.76°, respectively assigned to the (110), (220), and (310) planes of tetragonal perovskite structure [24], which indicated that the desired perovskite film was formed. Obviously, the main peak intensity of the solvent-annealed CH3NH3PbI3 film was stronger and sharper than that the as-prepared, which was consistent with the trend of FWHM and shown in Figure 3b. The average crystalline size of (110), (220), and (310) planes in the as-deposited and solventannealed film was estimated by resolving the characteristic peaks using Scherrer's formula

Results and Discussions
where K is Scherrer constant, D is the average grain size, B is the half-width of the measured diffraction peak, θ is the diffraction angle, and γ is the X-ray wavelength (0.154056 nm).   Figure 3 shows the X-ray diffraction (XRD) pattern and its corresponding FWHM and average crystal size of as-prepared and solvent-annealed CH3NH3PbI3 films. All the samples exhibited intense diffraction peaks at 14.02°, 28.32°, and 31.76°, respectively assigned to the (110), (220), and (310) planes of tetragonal perovskite structure [24], which indicated that the desired perovskite film was formed. Obviously, the main peak intensity of the solvent-annealed CH3NH3PbI3 film was stronger and sharper than that the as-prepared, which was consistent with the trend of FWHM and shown in Figure 3b. The average crystalline size of (110), (220), and (310) planes in the as-deposited and solventannealed film was estimated by resolving the characteristic peaks using Scherrer's formula where K is Scherrer constant, D is the average grain size, B is the half-width of the measured diffraction peak, θ is the diffraction angle, and γ is the X-ray wavelength (0.154056 nm). It can be seen from the Figure 3c that the average crystal size of solvent-annealed perovskite film became bigger. The more intense characteristic peaks and larger average crystal size both attributed to the improved crystallinity and/or composition homogeneity with fewer low-dimensional defects and less scattering between grain boundaries in the solvent-annealed perovskite film. The elemental ratio of Pb and I of the as-prepared and solvent-annealed CH3NH3PbI3 films measured by EDS is shown in Table 1. The Pb/I of as-prepared and solvent-annealed CH3NH3PbI3 films, respectively, were 1/3.04 and 1/3.03, which was in well agreement with the theoretical stoichiometry of 1/3. In order to further examine the compositional distribution throughout the It can be seen from the Figure 3c that the average crystal size of solvent-annealed perovskite film became bigger. The more intense characteristic peaks and larger average crystal size both attributed to the improved crystallinity and/or composition homogeneity with fewer low-dimensional defects and less scattering between grain boundaries in the solvent-annealed perovskite film.
The elemental ratio of Pb and I of the as-prepared and solvent-annealed CH 3 NH 3 PbI 3 films measured by EDS is shown in Table 1. The Pb/I of as-prepared and solvent-annealed CH 3 NH 3 PbI 3 films, respectively, were 1/3.04 and 1/3.03, which was in well agreement with the theoretical stoichiometry of 1/3. In order to further examine the compositional distribution throughout the solvent-annealed films, Figure 4 shows the elemental mapping of Pb and I with uniformly distributed, indicating that the solvent annealing is a reliable method to fabricate the desired perovskite film.  The elemental ratio of Pb and I of the as-prepared and solvent-annealed CH3NH3PbI3 films measured by EDS is shown in Table 1. The Pb/I of as-prepared and solvent-annealed CH3NH3PbI3 films, respectively, were 1/3.04 and 1/3.03, which was in well agreement with the theoretical stoichiometry of 1/3. In order to further examine the compositional distribution throughout the solvent-annealed films, Figure 4 shows the elemental mapping of Pb and I with uniformly distributed, indicating that the solvent annealing is a reliable method to fabricate the desired perovskite film.     Figure 5 shows the optical absorption spectra and the corresponding Tauc plots of the as-prepared and solvent-annealed MAPbI 3 films. The optical absorption of the as-prepared perovskite film was much lower than that of the solvent-annealed in the wavelength range from 300 nm to 750 nm because of more defect states caused by the smaller grains, as shown in the SEM images. Additionally, all the samples showed a sharp absorption edge in the absorption spectra at about 780 nm. In addition, the optical band-gap could be calculated according to the following equation where A is a constant that is uncorrelated with the value of the optical bandgap, hv is the photon energy, α is the optical absorption coefficient obtained from the measured absorption spectrum, and E g is the bandgap of the light absorbing material. The calculated optical band gap of the as-prepared and solvent-annealed CH 3 NH 3 PbI 3 films were 1.58 eV and 1.59 eV, respectively, which was consistent with the reported for MAPbI 3 [25]. Figure 6 shows the Photoluminescence (PL) spectra of CH 3 NH 3 PbI 3 films with and without solvent annealing. It can be observed that the PL peaks of CH 3 NH 3 PbI 3 films with and without solvent annealing were respectively located at 781 and 785 nm, which agreed with the above calculated bandgap result. In addition, the intensity of the PL peak of CH 3 NH 3 PbI 3 film with solvent annealing was incredibly stronger than that of as-prepared perovskite film, suggesting that the number of defects and trap states of CH 3 NH 3 PbI 3 film reduced after solvent annealing.
where is a constant that is uncorrelated with the value of the optical bandgap, ℎ is the photon energy, is the optical absorption coefficient obtained from the measured absorption spectrum, and is the bandgap of the light absorbing material. The calculated optical band gap of the as-prepared and solvent-annealed CH3NH3PbI3 films were 1.58 eV and 1.59 eV, respectively, which was consistent with the reported for MAPbI3 [25]. Figure 6 shows the Photoluminescence (PL) spectra of CH3NH3PbI3 films with and without solvent annealing. It can be observed that the PL peaks of CH3NH3PbI3 films with and without solvent annealing were respectively located at 781 and 785 nm, which agreed with the above calculated bandgap result. In addition, the intensity of the PL peak of CH3NH3PbI3 film with solvent annealing was incredibly stronger than that of as-prepared perovskite film, suggesting that the number of defects and trap states of CH3NH3PbI3 film reduced after solvent annealing.    Table 2. It could be inferred from the figure that the performance of solvent-annealed perovskite solar cell with Voc of where is a constant that is uncorrelated with the value of the optical bandgap, ℎ is the photon energy, is the optical absorption coefficient obtained from the measured absorption spectrum, and is the bandgap of the light absorbing material. The calculated optical band gap of the as-prepared and solvent-annealed CH3NH3PbI3 films were 1.58 eV and 1.59 eV, respectively, which was consistent with the reported for MAPbI3 [25]. Figure 6 shows the Photoluminescence (PL) spectra of CH3NH3PbI3 films with and without solvent annealing. It can be observed that the PL peaks of CH3NH3PbI3 films with and without solvent annealing were respectively located at 781 and 785 nm, which agreed with the above calculated bandgap result. In addition, the intensity of the PL peak of CH3NH3PbI3 film with solvent annealing was incredibly stronger than that of as-prepared perovskite film, suggesting that the number of defects and trap states of CH3NH3PbI3 film reduced after solvent annealing.    Table 2. It could be inferred from the figure that the performance of solvent-annealed perovskite solar cell with Voc of   Table 2. It could be inferred from the figure that the performance of solvent-annealed perovskite solar cell with V oc of 0.94 V, J sc of 19.97 mA/cm 2 , FF of 53%, and PCE of 9.92% was much better than that of as-prepared one with V oc of 0.77 V, J sc of 10.04 mA/cm 2 , FF of 34%, and PCE of 2.64%. Here, the reason why the fill factors (FF) of the devices were low may be the presence of the underestimated defects in the interface, especially between the absorption layer and the hole transport layer, leading to carrier recombination loss during electron transporting. Figure 8 shows the histograms of the PCE, V oc , J sc and FF distribution based on 19 perovskite solar cells. The devices based on the solvent-annealed films showed high reproducibility with average PCE of 8.98 ± 0.60%. The improved device performance and high reproducibility were contributed to the increased optical absorption, micron-sized crystal grain and fewer defects of CH 3 NH 3 PbI 3 film after solvent-annealing treatment, which indicated that the solvent annealing was a promising method to significantly improve the quality of perovskite film and enhanced the efficiency of CH 3 NH 3 PbI 3 solar cells.
films showed high reproducibility with average PCE of 8.98 ± 0.60%. The improved device performance and high reproducibility were contributed to the increased optical absorption, micronsized crystal grain and fewer defects of CH3NH3PbI3 film after solvent-annealing treatment, which indicated that the solvent annealing was a promising method to significantly improve the quality of perovskite film and enhanced the efficiency of CH3NH3PbI3 solar cells.

Conclusion
In this paper, after the solvent annealing process, larger grains, higher crystallinity, well-defined grain structure, desired element stoichiometric ratio, and excellent optical properties of CH3NH3PbI3 films prepared by single source evaporation can be obtained. Solar cell devices based on the perovskite films with solvent-annealing treatment achieved a much higher power conversion

Conclusions
In this paper, after the solvent annealing process, larger grains, higher crystallinity, well-defined grain structure, desired element stoichiometric ratio, and excellent optical properties of CH 3 NH 3 PbI 3 films prepared by single source evaporation can be obtained. Solar cell devices based on the perovskite films with solvent-annealing treatment achieved a much higher power conversion efficiency of 9.92% than that of 2.64% of as-prepared, suggesting that single-source vacuum thermal evaporation assembled with the solvent annealing was a potential method to prepare the high-quality perovskite films for photovoltaic application.