Constructing water-resistant CH3NH3PbI3 perovskite films via coordination interaction

Organic–inorganic halide CH3NH3PbI3 (MAPbI3) perovskite solar cells (PSCs) have attracted intensive attention due to their high power conversion efficiency and low fabrication cost. However, MAPbI3 is known to be very sensitive to humidity, and the intrinsic long-term stability of the MAPbI3 film remains a critical challenge. 2-Aminoethanethiol (2-AET) was used as a ligand to bridge the organic compound (MAI) and inorganic compound (PbI2), which restricted the fast growth of PbI2 to realize the synchronous growth environment of MAI and PbI2 crystals, resulting in the formation of a compact MAPbI3 film with polygonal grains. Due to the compact (PbI2)–2-AET–(MAI) molecule barrier layers in the MAPbI3 structure, the resulting perovskite films showed excellent intrinsic water-resistance, with the MAPbI3 perovskite crystal structure retained for a long time (>10 minutes) after immersion in water. This work makes a step towards obtaining long-term stable MAPbI3 perovskite devices.


Introduction
Methylammonium lead halide organic-inorganic hybrid perovskites, such as CH 3 NH 3 PbI 3 (MAPbI 3 ), have recently attracted enormous attention for thin-lm solar cells due to their high optical absorption coefficient, long free carrier diffusion length, low exciton binding energy and simple low temperature solution-based processing. [1][2][3][4] Such organic-inorganic hybrid perovskite materials were rst made several decades ago, the unique structure of the organic-inorganic perovskites show the potential applications in the electrical and optical elds. [5][6][7][8][9] To date, the highest power conversion efficiencies (PCEs) of the perovskite solar cells (PSCs) are above 21%, 10 approaching the record efficiencies of monocrystalline silicon-based solar cells (25.6%) and thin lm single-crystalline GaAs cells (28.8%). 11,12 A serious current deciency of PSCs is the high sensitivity of MAPbI 3 to humidity. The PCE drops nearly 90% under an ambient environment (T ¼ 25 C, 30-50% humidity) in a few days. 13 It has been shown that MAPbI 3 can be easily degraded to MAI, PbI 2 , and HI in a few hours under a high humidity environment. 14,15 Moisture in air is considered a key factor causing the decomposition of MAPbI 3 . Compared to monocrystalline silicon-based solar cells with 20-30 years guaranteed lifetime, the poor stability of PSCs is a crucial barrier to their practical applicability.
Moisture stability has become one of the focus areas of MAPbI 3 research. For example, the surfaces of devices or perovskite were covered by p-type metal oxides, hydrophobic materials or insulating polymers to enhance the stability of PSCs under an ambient atmosphere. 13,16,17 However, the intrinsic water sensitivity of perovskite lms remains an unsolved problem. Recently, it was demonstrated that hydrate intermediate compounds, such as MAPbI 3 $H 2 O and (MA) 4 -PbI 6 $2H 2 O, were formed at the initial stage of the MAPbI 3 decomposition process under controlled humidity conditions, and the degradation reaction could be reversed by drying treatment. [18][19][20] In MAPbI 3 perovskite crystals, inorganic PbI 2 frameworks and organic MA + cations are bound together by weak hydrogen and ionic bonds. The migration activation energy for MA + is approximately 0.84 eV, determined by rstprinciples calculations. [21][22][23] The strength of hydrogen bonding will be reduced when the perovskite crystal is exposed to H 2 O or other polar solvents, and MA + would readily separate from octahedral PbI 6 , resulting in a rapid decomposition of MAPbI 3 and degradation of solar cell performance. It has been suggested that chemical modication between organic cation MA + and inorganic framework PbI 6 in MAPbI 3 perovskite can be a way to achieve inherent moisture stability. [24][25][26] For example, butylphosphonic acid 4-ammonium chloride was chosen to crosslink neighboring perovskite grains through hydrogen bonding, leading to increased photovoltaic performance and moisture stability. 24 2-Aminoethanethiol (2-AET) has been extensively used as a bidentate chelating agent in coordination chemistry, and shows a high affinity toward binding to Pb 2+ through the thiolate group. [27][28][29][30][31] In addition, the ammonium group in 2-AET molecules would bind MAI by hydrogen bonds in the perovskite structure. Therefore, 2-AET can be a great ligand to improve the intrinsic long-term stability of MAPbI 3 perovskite.
The present paper reports the fabrication of a novel waterresistant MAPbI 3 $2-AET perovskite lm, with the MAPbI 3 crystal structure retained aer immersion in water at room temperature (RT) for a long time (>10 minutes). When 2-AET was added into perovskite precursor solution, the thiolate group of 2-AET molecules facilitates strong interaction with PbI 2 , while the ammonium group can anchor MAI to form PbI 2 $2-AET$MAI when DMF is removed, providing a homogeneous environment for perovskite crystal nucleation and growth, and leading to uniform perovskite lms with excellent crystallinity.

Experimental section
Materials synthesis and perovskite lm fabrication CH 3 NH 3 I (MAI) was prepared by the reported method. 32 Methylamine (33 wt% in methanol) was mixed with hydroiodic acid (HI, 57 wt%) in a molar ratio of 1.2 : 1 at 0 C. Aer stirring for 2 h under a N 2 atmosphere, the solvent was removed by using a rotary evaporator at 50 C. Then the obtained powder was washed with ethyl acetate (EA) three times, and then recrystallized in methanol. Finally, the product powder was dried at 60 C in a vacuum oven for 24 h.
45 mL standard perovskite precursor solution was spin coated onto FTO substrates at 3500 rpm for 20 s and 5000 rpm for 10 s, followed by annealing at 100 C for 10 min.
A MAI$2-AET complex was obtained by adding 3 mL ethyl acetate into 1 mL MAI/2-AET (1 : 1 mol%) DMF solution, and the solid complex was ltered and dried at 60 C in a vacuum oven for 3 h. The MAI/PbI 2 complex was obtained by adding 3 mL ethyl acetate into perovskite precursor solution (1 : 1 mol%), and the solid complex was ltered and dried at 60 C in a vacuum oven for 3 h. The MAI$2-AET$PbI 2 complex was obtained by adding 3 mL ethyl acetate into 1 mL 2-AET added perovskite precursor solution (1 : 1 : 1 mol%), and the solid complex was ltered and dried at 60 C in a vacuum oven for 3 h.

Characterization
The IR spectra were measured by using a Bruker Vertex 80v Fourier transform infrared (FTIR) spectrometer. The scanning electron microscopy (SEM) measurements were performed using a cold eld emission scanning electron microscope (SU8020, Hitachi). The crystal structure was analyzed by using an X-ray diffractometer (PANalytical, Netherlands) with a monochromatic Cu Ka radiation source (l ¼ 1.54056Å). The absorption spectra were recorded using an ultraviolet-visible (UV-vis) spectrophotometer (UV-3600, Shi-madzu). The photoluminescence (PL) spectra were measured using a steady state uorescence spectrometer (FLS980, Edinburgh), and the excitation wavelength was 465 nm. The weight loss was measured on a TGA/DSC instrument (STA 4493F3, NETZSCH) from 30-350 C with a heating rate of 5 C min À1 . Fig. 1(a) shows the chemical structures of MAI, PbI 2 and 2-AET; the 2-AET molecule contains thiolate and ammonium functional groups. In order to completely dissolve PbI 2 in DMF solvent, a heating process at 70 C was usually needed due to the weak interaction of Pb-O coordination bonds between PbI 2 and DMF. [34][35][36][37][38] However, as shown in Fig. S1, † PbI 2 was quickly dissolved in DMF when 0.2 M 2-AET was added into DMF solution at RT. Fig. 1(b) shows the Fourier transform infrared (FTIR) spectra of pure DMF, PbI 2 and PbI 2 /2-AET DMF solution. The stretching vibration of the C]O bond appears at 1660 cm À1 for pure DMF, which is shied to 1633 cm À1 for PbI 2 DMF solution, and then the shoulder on the shorter wavenumber side disappears when 2-AET is added into PbI 2 solution. Considering that the interaction between Pb-S coordination bonds is stronger than that between Pb-O coordination bonds, 39-41 the Pb-O coordination bond is replaced by the Pb-S coordination bond when 2-AET is added into PbI 2 DMF solution. It demonstrates that the strong interaction of the Pb-S coordination bond between PbI 2 and 2-AET is formed in PbI 2 DMF solution. solution, and there is no obvious peak shi, which implies that the interaction between MAI and 2-AET is not formed in DMF solution. In order to further understand the effect of 2-AET on the interaction among PbI 2 , MAI and DMF, the solid complexes were extracted from the precursor solution with ethyl acetate (EA) as an extraction agent. Fig. S3(a) † shows the complex without 2-AET dividing into a white upper layer and a yellow bottom layer. The white and yellow products are MAI and PbI 2 , respectively. It demonstrates that there is weak interaction among PbI 2 , MAI and DMF. No separated layers in the small volume of yellow complexes are found when 2-AET is added into the precursor solution. The complex is uid, indicating that the strong interaction among PbI 2 , MAI and 2-AET is formed aer extraction. Fig. S3(b) † shows the Tyndall effect in the complex with the addition of 2-AET under red light illumination, indicating the colloidal characteristic of the complex. Fig. 1(c) and (d) show FTIR spectra of MAI, 2-AET and MAI$2-AET solid complexes. The N-H bending vibration and N-H stretching vibration appear at 1597 cm À1 and 2989 cm À1 for 2-AET, and at 1564 cm À1 and 3099 cm À1 for MAI. However, the N-H bending vibration and N-H stretching vibration of the MAI$2-AET complex shi to 1578 cm À1 and 3059 cm À1 , respectively. The N-H vibration frequency of the MAI$2-AET complex is between the 2-AET and MAI, providing evidence for interaction between 2-AET and MAI. Fig. S3(c) and (d) † show X-ray diffraction (XRD) patterns of the complexes with and without the additive of 2-AET. It can be seen that the diffraction peaks could not be indexed to MAI, PbI 2 or MAPbI 3 for the complex without 2-AET, as reported in the literature. 42 XRD patterns of the complex with the addition of 2-AET show several peaks of the new intermediate phase, due to the coordination interaction of 2-AET. Notably, the color of the solid complex without 2-AET changes from yellow to black at the edge aer measurement, but the solid complex with the 2-AET sample remains yellow, which implies that 2-AET can retard the change of the intermediate phase to the MAPbI 3 phase at RT. The intermediate phases changed to the MAPbI 3 phase when the complexes were annealed at 100 C for 30 min in air. The thiolate group in 2-AET can interact with PbI 2 to form a PbI 2 $2-AET complex in 2-AET added perovskite precursor solution, while the ammonium group can anchor MAI to form PbI 2 $2-AET$MAI when DMF is removed.

Results and discussion
The scanning electron microscopy (SEM) images, XRD patterns and statistical grain size distribution of nal MAPbI 3 $(x) 2-AET perovskite lms are illustrated in Fig. 2. In the case of the perovskite lm without 2-AET, it clearly shows the typical branchlike perovskite grains with poor coverage on the FTO substrate, which is in accordance with the literature. 43,44 The morphology of MAPbI 3 perovskite is usually related to the reaction rate between PbI 2 and MAI, which depends on their respective solubility and diffusivity in precursor solution. 41,45,46 Due to the good crystallinity and poor solubility of PbI 2 in precursor solution, 35 PbI 2 crystals rst precipitate out to form an inorganic framework during the spin coating process, and then react with MAI to form MAPbI 3 crystals. So the MAPbI 3 crystallization process is dominated by the crystallization rate of PbI 2 . The asynchronous reaction for PbI 2 and MAI is likely to lead to the growth of branch-like crystals of MAPbI 3 . Retarding the crystallization rate of PbI 2 is considered an efficient method to obtain uniaxial grains for high performance of PSCs. 38,41 As shown in Fig. 2 No obvious position shi of diffraction peaks is observed. A uniform and compact perovskite lm with excellent crystallinity is one of the essential requirements for PSC application. [45][46][47]49 The effect of thermal treatment on the MAPbI 3 perovskite lms was investigated. Fig. 3  coverage of the MAPbI 3 $(0.75)2-AET perovskite lm is higher than that of the MAPbI 3 $(0.5)2-AET perovskite lm aer annealing, indicating that the growth rate of perovskite is accelerated by 2-AET. Fig. 3(g) and (h) show XRD patterns of MAPbI 3 perovskite lms aer thermal annealing at 100 C for different durations. All annealed perovskite lms present increased intensity of diffraction peaks of (110) and (220) crystal planes of the MAPbI 3 perovskite structure at 14.14 and 28.45 . There is no detectable peak in the lms with increasing the duration of thermal annealing. Table S1 † shows the decreased full width at half maximum (FWHM) from 0.22 to 0.11, demonstrating the increased crystallinity of perovskite with increased annealing time. The characteristic peaks of the perovskite lms are found to shi to lower angles aer annealing for 3 min, and then shi back to higher angles aer annealing for 12 min. The reason is ascribed to the effect of 2-AET in the MAPbI 3 structure. For MAPbI 3 $(0.5)2-AET, a part of the free PbI 2 and MAI change to MAPbI 3 aer spin coating. The initially formed lm has a poor crystallinity and weak optical absorption of MAPbI 3 as shown in Table S1 and Fig. S5. † During the annealing process, the 2-AET additive was gradually pushed out to the boundaries of perovskite grains, which increased the perovskite crystal plane spacing. Subsequently, perovskite grains grew up, which tended to gradually decrease the crystal plane spacing. Fig. 4 is the schematic illustration of the crystallization process of MAPbI 3 . The branch-like crystals of MAPbI 3 are obtained from standard precursor solution via the asynchronous reaction between PbI 2 and MAI. When 2-AET is added into the standard precursor solution, the PbI 2 $2-AET coordination complex is formed to retard the fast crystallization of PbI 2 . Aer spin coating at RT, the PbI 2 $MAI$DMF$2-AET lms are obtained, which provide the synchronous growth environment of MAI and PbI 2 crystals. There are three stages to nish the crystallization of perovskite crystals during the annealing process. The remnant DMF gradually evaporates and the MAPbI 3 $PbI 2 $MAI$2-AET intermediate phase is formed during the annealing stage (1). Then the intermediate phases obtain enough thermal energy to break the coordination interaction of 2-AET in the annealing stage (2), and the 2-AET additive is concentrated in the grain boundaries of perovskite grains. Finally, together with the growth process of perovskite grains, the 2-AET additive is driven from grain boundaries to the surface of perovskite grains in the stage (3) as shown in Fig. S4(d). † Fig. 5(a) shows UV-vis absorbance of MAPbI 3 perovskite lms prepared with different additions of 2-AET. Without 2-AET, the shape of the spectrum shows clear evidence that the lm does not provide perfect coverage and consists of areas of high optical density mixed with uncovered parts. 50 The perovskite lm prepared with 0.15 M 2-AET shows less of this effect. Clearly, the higher concentration of M 2-AET leads to a highquality lm with close to perfect coverage. We point out that the absorption of perovskite lms gradually drops when the addition is more than 0.25 M. It should be attributed to the strong coordination interaction of 2-AET in the perovskite lm, which forms the complex in the nal lm, as indicated by the unknown XRD peaks in Fig. S6. † Besides, the decrease of the  absorption of the perovskite lms in the 775-800 nm region is caused by the reduced light scattering. The weak light scattering for 2-AET added perovskite lms is attributed to the uniform morphology. UV-vis absorption spectra also show a blue shi by $30 nm from MAPbI 3 to MAPbI 3 $(0.25)2-AET, which is another evidence for the interaction between PbI 2 , MAI and 2-AET. The steady-state photoluminescence (PL) spectra in Fig. 5(b) show that the peak of the MAPbI 3 $(0.5)2-AET perovskite lm is blue shied by $30 nm compared to MAPbI 3 perovskite. Fig. 6(a) and (b) show the photographs of MAPbI 3 and MAPbI 3 $(0.5)2-AET perovskite lms immersed in water for different times at RT. SEM images of the lms are shown in Fig. S7 and S8. † It can be seen that the color of MAPbI 3 $(0.5) 2-AET perovskite lms remains dark brown aer immersion in water, while MAPbI 3 perovskite lms undergo a rapid color change from dark brown to yellow. This process has also been recorded in Movie S1. † XRD patterns of MAPbI 3 and MAPbI 3 $(0.5) 2-AET are shown in Fig. 6(c) and (d). For MAPbI 3 perovskite lms, PbI 2 characteristic peaks at 12.7 , 25.9 , and 34.3 appear aer immersion in water for 10 s, and are well indexed to hexagonal PbI 2 (JCPDS card no. 07-0235). For MAPbI 3 $(0.5)2-AET perovskite lms, characteristic peaks at 14.15 and 28.46 of MAPbI 3 remain aer immersion in water from 10 s to 300 s, without the presence of detectable crystalline PbI 2 . However, the intensity of the peaks decreases gradually with increasing immersion time from 10 s to 300 s. No visible diffraction peak offset nor impurity peaks were observed. Compared to the pristine MAPbI 3 $(0.5)2-AET perovskite lm, the UV-vis absorption of the lm is rst enhanced in the wavelength range from 450 nm to 800 nm aer immersion in water for 10 s as shown in Fig. 6(e). The possible reason is that the excess 2-AET in the perovskite grain boundary is gradually dissolved in water to form holes in the perovskite lm (as shown in Fig. S8 †), which enhances the light trapping and the light scattering. Then the absorption is gradually decreased in the short wavelength range with increasing immersion time. However, the absorption edge of PbI 2 at $520 nm as reported in the literature is not found. 51 A red shi of the absorption edge from $770 nm to $780 nm was also observed for all MAPbI 3 $(0.5)2-AET perovskite lms aer immersion due to the light scattering. As shown in Fig. 6(f), aer immersion, the PL peaks also shi 10 nm from 742 nm to 752 nm. Aer that, the PL curves do not change with increasing immersion time, indicating that the structure of MAPbI 3 $(0.5)2-AET is stable in water. Fig. 7(a) shows XRD patterns of MAPbI 3 perovskite lms prepared with low addition of 2-AET aer immersion in water for 15 s. The degradation rate of MAPbI 3 $(x)2-AET increases as the amount of 2-AET addition decreases. As shown in Table S2, † aer immersion in water for 15 s, the diffraction peak of PbI 2 at 12.79 shis to 12.22 when the addition of 2-AET increases from 0 M to 0.3 M. The decreased intensity with increased FWHM of PbI 2 characteristic peaks implies that the MAPbI 3 degradation process is suppressed by the presence of 2-AET. The schematic diagram of the degradation process of MAPbI 3 and possible mechanism of water-resistant MAPbI 3 $(x)2-AET perovskite in water are presented in Fig. 7(b) and (c). The   MAPbI 3 lm prepared without the addition of 2-AET degrades rapidly to hexagonal PbI 2 . However, in the case of the MAPbI 3 lm prepared with addition of 2-AET, the distribution of 2-AET in perovskite grain boundaries forms compact (PbI 2 )-2-AET-(MAI) molecule barrier layers, which can effectively prevent H 2 O from penetrating into MAPbI 3 crystals. As a result, although 2-AET dissolves easily in water, MAPbI 3 $(x)2-AET perovskite shows excellent water-resistance due to the coordination interaction between 2-AET and MAPbI 3 by hydrogen bonds. Fig. 8 shows differential scanning calorimetry/thermogravimetric analysis (DSC/TGA) curves of MAPbI 3 and MAPbI 3 $(0.75) 2-AET. The color of MAPbI 3 $(0.75)2-AET powders changed from yellow to black aer the TGA/DSC measurement, while the color of MAPbI 3 still remained black. The weight loss for both perovskite powders is approximately the same, 4% from RT to 125 C, which is attributed to the evaporation of physically adsorbed H 2 O or DMF. For MAPbI 3 powder, the TGA curve stepwise decreases with an obvious endothermic peak at $125 C, likely due to the removal of combined water. 52 In the case of MAPbI 3 $(0.75)2-AET powder, the mass consecutively drops with an increasing temperature from 30 C to 125 C, and there is no endothermic peak at $125 C, so the weight loss is related to evaporation of DMF. This may imply that the reported hydrate intermediate compounds (MAPbI 3 $H 2 O or (MA) 4 PbI 6 $2H 2 O) are not formed in MAPbI 3 $(0.75)2-AET powder. [18][19][20] No visible weight loss is observed between 125 C and 225 C for MAPbI 3 powder, which is consistent with other reports. [52][53][54] The weight loss is 6.64% between 150 C and 225 C for MAPbI 3 $(0.75)2-AET powder, which is close to the amount of (8.54%) 2-AET added into precursor solution. However, as shown in Fig. S9, † for pure 2-AET powder, the weight loss is only 6.27%. In addition, the DSC curve shows obvious endothermic peaks at 200-225 C, so the poor thermal stability should be attributed to a series of reactions between MAPbI 3 and 2-AET, and the reaction products will be evaporated from the perovskite powder.

Conclusions
Excellent and intrinsic water-resistant MAPbI 3 $(x)2-AET perovskite lms have been grown and demonstrated. 2-AET is a ligand with strong coordination interactions not only observed in the perovskite precursor solutions, but also in the resulting perovskite crystal structure. The colloidal PbI 2 $2-AET$MAI complexes have a great inuence on the nucleation and growth process, and the synchronous growth of MAI and PbI 2 crystals is realized during a one step spin coating process. MAPbI 3 $(x)2-AET during the thermal annealing process leads to the growth of compact perovskite lms with signicant preferential orientation of (110) and (220) planes and enhanced crystallinity. Due to the compact (PbI 2 )-2-AET-(MAI) molecule barrier layers in the MAPbI 3 structure, the perovskite lms show excellent intrinsic water-resistance, with the MAPbI 3 perovskite crystal structure retained aer immersion in water for a long time (>10 minutes).