Modified Sequential Deposition Route through Localized-Liquid-Liquid-Diffusion for Improved Perovskite Multi-Crystalline Thin Films with Micrometer-Scaled Grains for Solar Cells

High-class perovskite film with beautiful surface morphology (such as large-size grain, low defect density, good continuity and flatness) is normally believed to be a very important factor for high-efficiency perovskite solar cells (PSCs). Here, we report a modified sequential deposition route through localized-liquid-liquid-diffusion (LLLD) for qualified perovskite multi-crystalline thin films with micrometer-scaled grains for solar cells. We adopted a contact-type drop method to drop Methylammonium iodide (MAI) solution and have successfully used high-concentration MAI solution (73 mg/mL) to transform PbI2 film into high-class perovskite film via our route. A high efficiency of 10.7% was achieved for the device with spongy carbon film deposited on a separated FTO-substrate as a counter electrode under one sun illumination, which is the highest efficiency (as 2.5 times as previous efficiency) ever recorded in perovskite solar cells with a such spongy carbon/FTO composite counter electrode. The preparation techniques of high-class perovskite thin films under ambient conditions and the cheap spongy carbon/FTO composite counter electrode are beneficial for large-scale applications and commercialization.


Introduction
In recent years, perovskite solar cells (PSCs) have been recognized as the most promising alternative to conventional photovoltaic devices due to their high efficiency and simple process [1][2][3]. High-class perovskite thin film, with beautiful surface morphology (such as large-size grain, low defect density, good continuity, and flatness), is normally believed to be a very important factor for high-efficiency perovskite solar cells [4][5][6][7][8]. M. Grätzel et al. firstly developed the sequential deposition method (so-called two-step method) to prepare CH 3 NH 3 PbI 3 perovskite thin films [9]. Since then, a lot of effort has been paid to developing the two-step method, similar to the two-step spin-coating process [10][11][12], and Xu et al. have demonstrated that the two-step spin coating method enables perovskite layer morphology to control and easily fabricate products [13]. Previous researchers prepared perovskite films via the conventional two-step coating method with a necessary step of PbI 2 layer annealing [9][10][11][12][13]. In fact, we found the properties of the perovskite thin film are very sensitive to the annealing time of PbI 2 layer especially under the dry ambient conditions used in our experiments. film are very sensitive to the annealing time of PbI2 layer especially under the dry ambient conditions used in our experiments.
Xu et al.'s previous study on the effect of humidity on the crystallization of two-step spin-coated perovskites suggested that appropriate moisture can facilitate the reaction between PbI2 and MAX (X = I, Cl), whereas the PbI2 film reacts with MAX slowly a under dry inert atmosphere [14]. H2O helps MAI to penetrate into the thick PbI2 to form thick film with a pure MAPbI3 phase and produce bigger gains by slowing down the perovskite crystallization rate [15]. However, the dry atmosphere (~10% relative humidity) was our experimental condition.
In order to explore the effect of the annealing time of PbI2 thin film on properties of corresponding perovskite thin film, we annealed PbI2 film at 70 °C for 0 s, 10 s, and 20 s, respectively. We found that the 0 s annealing PbI2 film was more conducive to the formation of high-class perovskite films in our experiment, which means that the sequential deposition route through LLLD is more conducive to the formation of qualified perovskite multi-crystalline films. In addition, we adopted a contact-type drop method to drop MAI solution in order to reduce the damage of MAI solution to PbI2 thin liquid film (illustrated in Figure 1). Previously, if MAI concentration is too high (≥70 mg/mL), most of the PbI2 precursor gets dissolved and exists as PbI4 2− in the solution, and only very little can reprecipitate to form MAPbI3 nanostructures [16]. Im et al. found that the size of the MAPbI3 cuboids increases with decreasing CH3NH3I concentration [17]. However, the MAI concentration in our experiment was 73 mg/mL and the corresponding MAI solution also successfully transformed PbI2 film into high-class perovskite film via LLLD. A high-class perovskite film with micrometer-scaled grain, low defect density, densification, and flatness was obtained by our process.
The tune of the formation process and composition of the light-absorber layer in PSCs has contributed to obtaining certified power conversion efficiencies (PCE) > 20% [18,19]. These PCEs have been obtained while perovskite solar cells were prepared using the single-step method (in combination with solvent engineering) or sequential deposition method or vapor deposition with an important step of device fabrication being the thermal evaporation of metals to form the counter electrode (CE). However, the cost of these metallic CEs is prohibitively high for large-scale applications and commercialization.
Up to now, some researchers have developed carbon CEs to cut the costs of perovskite devices [20][21][22][23]. Yang et al. prepared a 2.6%-efficient perovskite solar cell with candle soot film deposited on a separated substrate as the carbon/FTO composite counter electrode [22]. In our research group, previously, a spongy carbon/FTO composite structure was adopted as a counter electrode and the corresponding cell achieved 4.24% power conversion efficiency (PCE) [23]. The perovskite solar cell was prepared via a modified sequential deposition route through LLLD under dry ambient conditions and got a 10.7% PCE in the size of 0.2 cm 2 , which is the highest efficiency ever recorded in perovskite solar cells with such a spongy carbon/FTO composite counter electrode. Previously, if MAI concentration is too high (≥70 mg/mL), most of the PbI 2 precursor gets dissolved and exists as PbI 4 2− in the solution, and only very little can reprecipitate to form MAPbI 3 nanostructures [16]. Im et al. found that the size of the MAPbI 3 cuboids increases with decreasing CH 3 NH 3 I concentration [17]. However, the MAI concentration in our experiment was 73 mg/mL and the corresponding MAI solution also successfully transformed PbI 2 film into high-class perovskite film via LLLD. A high-class perovskite film with micrometer-scaled grain, low defect density, densification, and flatness was obtained by our process. The tune of the formation process and composition of the light-absorber layer in PSCs has contributed to obtaining certified power conversion efficiencies (PCE) > 20% [18,19]. These PCEs have been obtained while perovskite solar cells were prepared using the single-step method (in combination with solvent engineering) or sequential deposition method or vapor deposition with an important step of device fabrication being the thermal evaporation of metals to form the counter electrode (CE). However, the cost of these metallic CEs is prohibitively high for large-scale applications and commercialization.
Up to now, some researchers have developed carbon CEs to cut the costs of perovskite devices [20][21][22][23]. Yang et al. prepared a 2.6%-efficient perovskite solar cell with candle soot film deposited on a separated substrate as the carbon/FTO composite counter electrode [22]. In our research group, previously, a spongy carbon/FTO composite structure was adopted as a counter electrode and the corresponding cell achieved 4.24% power conversion efficiency (PCE) [23]. The perovskite solar cell was prepared via a modified sequential deposition route through LLLD under dry ambient conditions and got a 10.7% PCE in the size of 0.2 cm 2 , which is the highest efficiency ever recorded in perovskite solar cells with such a spongy carbon/FTO composite counter electrode.

Device Fabrication
Fluorine-doped SnO 2 (FTO) substrates were ultrasonically cleaned with mixed solution (detergent and deionized water), glass water (acetone:deionized water:2-propanol = 1:1:1), and alcohol, respectively. Before using, the FTO was cleaned by Ultraviolet Ozone (UVO) for 90 min. Compact TiO 2 layer (bl-TiO 2 ) was deposited on FTO substrate by spin-coating an acidic titanium dioxide solution at 2000 rpm for 60 s. Then the substrate was annealed on a hotplate at 100 • C for 10 min, and sintered in a muffle furnace at 500 • C for 30 min. Mesoporous TiO 2 film (mp-TiO 2 ) was coated on bl-TiO 2 thin film by spin coating using a commercial 18NR-T TiO 2 paste (4%, solid content) at 2000 rpm for 30 s. Subsequently, the substrate was dried on a hotplate at 100 • C for 10 min, and then sintered in a muffle furnace at 500 • C for 60 min. After that, 599.3 mg of PbI 2 was dissolved in a component solvent (DMF:DMSO = 9.5:0.5) and then spin-coated onto a TiO 2 (bl-TiO 2 and mp-TiO 2 ) layer at 1500 rpm for 30 s, and then annealed on a hotplate at 70 • C for 0 s, 10 s, and 20 s, respectively.
After the PbI 2 film had cooled down to room temperature, the MAI solution (73 mg in 1 mL IPA) was spin coated onto the PbI 2 layer at 1500 rpm for 30 s (without loading time), and then a thermal annealing of 150 • C for 15 min under low humidity air condition (~10% humidity) was processed. The hole transporting layer was deposited on top of the perovskite layer by 3000 rpm for 30 s using a commercial 2,2 ,7,7 -tetrakis(N,N-dip-methoxyphenylamine)-9,9 -spirobifluorene (Spiro-OMeTAD) solution. Finally, some cleaned FTO glasses were used as substrates to obtain the soot of a burning candle as spongy carbon CEs. The spongy carbon films on FTO glasses were then pressed on the as-prepared uncompleted devices.

Characterization
An X-ray diffractometer (XRD) (Broker, D8 Focus, Beijing, China) was used to obtain XRD spectra from samples of perovskite films deposited on mp-TiO 2 /bl-TiO 2 /FTO substrates. The morphologies of these perovskite films were watched by a scanning electron microscope (SEM) (Zeiss SIGMA). The as-prepared devices were measured under full sun illumination (AM 1.5, 100 mW/cm 2 ).

Results and Discussion
In order to reduce the possible damage of the MAI solution to PbI 2 thin liquid film, we adopted a contact-type drop method to drop MAI solution (illustrated in Figure 1a), and the resulting perovskite films are shown in Figure 1b. In contrast to the non-contact-type drop method, this drop method reduced the damage of the MAI solution to PbI 2 thin film according to Figure 1b.
The perovskite films were prepared by sequential deposition route with PbI 2 0 s, 10 s, and 20 s annealing, which are named as sample A, sample B, and sample C, respectively. Scanning electron microscopy (SEM) images and image of perovskite layers prepared by two-step spin-coating with different PbI 2 annealing times are shown in Figure 2. We found that the surfaces of the perovskite layers gradually became a white color with the increase of PbI 2 annealing time, according to the image (Figure 2a) of the perovskite layers. From the surface SEM image (Figure 2b-d), it can be found that sample A showed obvious densification, sample B and C have double layers, and the bottom layer is dense and the top layer is loose. PbI 2 solution film reacts with MAI solution to form a high-class perovskite thin film with beautiful surface morphology (such as large-size grain, low defect density, good continuity, and flatness) via LLLD, whereas, PbI 2 solid film reacts with MAI solution to form loose perovskite thin film via localized-solid-liquid-diffusion (LSLD). The bottom layers of sample B and C are observed to contain large-size grain, but the top layer of sample B shows some small square holes and the top layer of sample C shows massive square unordered holes.  In the cross-sectional SEM image (Figure 2e) of sample A, we can see the FTO layer, bl-TiO2 layer, mp-TiO2 layer and perovskite layer clearly from the bottom to the top, and the longitudinal size of perovskite grains equal to the film thickness. This qualified perovskite film (sample A) indicates that the modified sequential deposition route through LLLD is beneficial to the growth of perovskite crystals under dry ambient condition and enhancement of the corresponding photovoltaic performance. Figure 3 shows XRD pattern of the samples prepared by two-step spin-coating method with different PbI2 annealing times. According to the Figure 3, there are no peaks of PbI2 (corresponding to ~26°) and MAI (corresponding to ~19°), which means there are no residuals of PbI2 and MAI in the samples, and the compositions of perovskites in the samples are very pure. As the PbI2 annealing time increased to 10 s, the (004) peak of the tetragonal structure for CH3NH3PbI3 emerged, which indicates a phase transition from cubic to tetragonal structure [24]. This result suggest that the perovskite in sample A and bottom layers of sample B and C is cubic structure, and the perovskite in top layers of sample B and C is tetragonal structure. In addition, the intensity ratio (~2.06) of the highest peak (110) of sample A to its secondary peak (220) is lightly above that of others (~1.93 and ~2.01). The above points means the crystal structure of sample A is better than others and sequential deposition route through LLLD is more propitious to the formation of qualified perovskite multi-crystalline films, which can enhance the corresponding device efficiency. In the cross-sectional SEM image (Figure 2e) of sample A, we can see the FTO layer, bl-TiO 2 layer, mp-TiO 2 layer and perovskite layer clearly from the bottom to the top, and the longitudinal size of perovskite grains equal to the film thickness. This qualified perovskite film (sample A) indicates that the modified sequential deposition route through LLLD is beneficial to the growth of perovskite crystals under dry ambient condition and enhancement of the corresponding photovoltaic performance. Figure 3 shows XRD pattern of the samples prepared by two-step spin-coating method with different PbI 2 annealing times. According to the Figure 3, there are no peaks of PbI 2 (corresponding to~26 • ) and MAI (corresponding to~19 • ), which means there are no residuals of PbI 2 and MAI in the samples, and the compositions of perovskites in the samples are very pure. As the PbI 2 annealing time increased to 10 s, the (004) peak of the tetragonal structure for CH 3 NH 3 PbI 3 emerged, which indicates a phase transition from cubic to tetragonal structure [24]. This result suggest that the perovskite in sample A and bottom layers of sample B and C is cubic structure, and the perovskite in top layers of sample B and C is tetragonal structure. In addition, the intensity ratio (~2.06) of the highest peak (110) of sample A to its secondary peak (220) is lightly above that of others (~1.93 and~2.01). The above points means the crystal structure of sample A is better than others and sequential deposition route through LLLD is more propitious to the formation of qualified perovskite multi-crystalline films, which can enhance the corresponding device efficiency. Schematic illustrations for LLLD and LSLD for qualified perovskite multi-crystalline film are shown in Figure 4a. As PbI2 annealing time increases, its solid part and the ratio of LLLD with MAI solution decreases. The changes in the morphology of the perovskite have finally happened. Due to the 0 s annealing PbI2 film was the solution state, and MAI could more easily diffuse to the PbI2 thin film interior via LLLD and transform PbI2 film into a high-class cubic structure perovskite film. Moreover, the PbI2 thin film had some of the solvent of DMF, which could facilitate perovskite conversion, improve the film morphology, and reduce crystal defects, and thus enhance charge transfer efficiency [25]. Due to the annealing time being relatively short, there are double layers in the 10 s annealing PbI2 film and 20 s annealing PbI2 film. The top layer is solid state, as a result of contacting with air during the spin-coating. MAI is difficult to diffuse to the solid top layer of the PbI2 thin film interior via LSLD and transform into poor perovskite film with a tetragonal structure. The bottom layer is in solution state for having not made contact with air during the spin-coating, which is similar to sample A, and can be transformed into a high-class cubic structure perovskite film. The hydrolysis of perovskite must be reduced because it was prepared in dry air. A combination of the above factors lead to the high-class perovskite films being fabricated by the sequential deposition route through LLLD under dry ambient conditions and the structure of the corresponding devices with the low-cost spongy carbon/FTO composite counter electrode is shown in Figure 4b.   Figure 5a,b show the light current-voltage (J-V) curves from reverse scan (RS) for the devices with a carbon/FTO composite electrode and gold electrode, measured under simulated sunlight with an intensity of 100 mW/cm 2 (AM 1.5 G). According to the J-V curve, it is found that the device (corresponding to PbI2 0 s annealing) via the modified sequential deposition route through LLLD Schematic illustrations for LLLD and LSLD for qualified perovskite multi-crystalline film are shown in Figure 4a. As PbI 2 annealing time increases, its solid part and the ratio of LLLD with MAI solution decreases. The changes in the morphology of the perovskite have finally happened. Due to the 0 s annealing PbI 2 film was the solution state, and MAI could more easily diffuse to the PbI 2 thin film interior via LLLD and transform PbI 2 film into a high-class cubic structure perovskite film. Moreover, the PbI 2 thin film had some of the solvent of DMF, which could facilitate perovskite conversion, improve the film morphology, and reduce crystal defects, and thus enhance charge transfer efficiency [25]. Due to the annealing time being relatively short, there are double layers in the 10 s annealing PbI 2 film and 20 s annealing PbI 2 film. The top layer is solid state, as a result of contacting with air during the spin-coating. MAI is difficult to diffuse to the solid top layer of the PbI 2 thin film interior via LSLD and transform into poor perovskite film with a tetragonal structure. The bottom layer is in solution state for having not made contact with air during the spin-coating, which is similar to sample A, and can be transformed into a high-class cubic structure perovskite film. The hydrolysis of perovskite must be reduced because it was prepared in dry air. A combination of the above factors lead to the high-class perovskite films being fabricated by the sequential deposition route through LLLD under dry ambient conditions and the structure of the corresponding devices with the low-cost spongy carbon/FTO composite counter electrode is shown in Figure 4b. Schematic illustrations for LLLD and LSLD for qualified perovskite multi-crystalline film are shown in Figure 4a. As PbI2 annealing time increases, its solid part and the ratio of LLLD with MAI solution decreases. The changes in the morphology of the perovskite have finally happened. Due to the 0 s annealing PbI2 film was the solution state, and MAI could more easily diffuse to the PbI2 thin film interior via LLLD and transform PbI2 film into a high-class cubic structure perovskite film. Moreover, the PbI2 thin film had some of the solvent of DMF, which could facilitate perovskite conversion, improve the film morphology, and reduce crystal defects, and thus enhance charge transfer efficiency [25]. Due to the annealing time being relatively short, there are double layers in the 10 s annealing PbI2 film and 20 s annealing PbI2 film. The top layer is solid state, as a result of contacting with air during the spin-coating. MAI is difficult to diffuse to the solid top layer of the PbI2 thin film interior via LSLD and transform into poor perovskite film with a tetragonal structure. The bottom layer is in solution state for having not made contact with air during the spin-coating, which is similar to sample A, and can be transformed into a high-class cubic structure perovskite film. The hydrolysis of perovskite must be reduced because it was prepared in dry air. A combination of the above factors lead to the high-class perovskite films being fabricated by the sequential deposition route through LLLD under dry ambient conditions and the structure of the corresponding devices with the low-cost spongy carbon/FTO composite counter electrode is shown in Figure 4b.   Figure 5a,b show the light current-voltage (J-V) curves from reverse scan (RS) for the devices with a carbon/FTO composite electrode and gold electrode, measured under simulated sunlight with an intensity of 100 mW/cm 2 (AM 1.5 G). According to the J-V curve, it is found that the device (corresponding to PbI2 0 s annealing) via the modified sequential deposition route through LLLD  Figure 5a,b show the light current-voltage (J-V) curves from reverse scan (RS) for the devices with a carbon/FTO composite electrode and gold electrode, measured under simulated sunlight with an intensity of 100 mW/cm 2 (AM 1.5 G). According to the J-V curve, it is found that the device (corresponding to PbI 2 0 s annealing) via the modified sequential deposition route through LLLD showed the best performance in major parameters (such as V oc , J sc and PCE, and their detailed values can be seen in Table 1) and the others showed obvious attenuation in major parameters. To be sure of the efficacy of our new coating processes and the reliability of current density data, we measured the IPCE of the devices with a carbon/FTO composite electrode, and the results are shown in Figure 5c. Generally, the device performance is increased with the decrease of PbI 2 annealing times, indicating that the sequential deposition route through LLLD is more conducive to the formation of qualified perovskite multi-crystalline films under dry ambient conditions. showed the best performance in major parameters (such as Voc, Jsc and PCE, and their detailed values can be seen in Table 1) and the others showed obvious attenuation in major parameters. To be sure of the efficacy of our new coating processes and the reliability of current density data, we measured the IPCE of the devices with a carbon/FTO composite electrode, and the results are shown in Figure  5c. Generally, the device performance is increased with the decrease of PbI2 annealing times, indicating that the sequential deposition route through LLLD is more conducive to the formation of qualified perovskite multi-crystalline films under dry ambient conditions.  According to the data of Table 1, it can be found that the photovoltaic performance of contact-type processed PSCs is better than the non-contact-type processed PSCs. This further validates the reduced damage of the MAI solution to the PbI2 thin film when dropping the MAI solution via contact-type method. A high efficiency of 10.7% was achieved for the device with the carbon/FTO composite electrode. Although this efficiency is not the highest in PSCs, it is the highest efficiency (as 2.5 times as previous efficiency) ever recorded in PSCs with such a spongy carbon/FTO composite counter electrode. Furthermore, the preparation techniques of high-class perovskite thin films and the cheap spongy carbon/FTO composite counter electrode, under absolute ambient conditions, are beneficial for large-scale applications and commercialization. According to the data of Table 1, it can be found that the photovoltaic performance of contact-type processed PSCs is better than the non-contact-type processed PSCs. This further validates the reduced damage of the MAI solution to the PbI 2 thin film when dropping the MAI solution via contact-type method. A high efficiency of 10.7% was achieved for the device with the carbon/FTO composite electrode. Although this efficiency is not the highest in PSCs, it is the highest efficiency (as 2.5 times as previous efficiency) ever recorded in PSCs with such a spongy carbon/FTO composite counter electrode. Furthermore, the preparation techniques of high-class perovskite thin films and the cheap spongy carbon/FTO composite counter electrode, under absolute ambient conditions, are beneficial for large-scale applications and commercialization.

Conclusions
A modified sequential deposition route through LLLD was more conducive to the formation of qualified perovskite multi-crystalline thin films with micrometer-scaled grain, low defect density, densification, and flatness under dry ambient condition. The two diffusion mechanisms of LLLD and LSLD were suggested and preliminarily confirmed. A contact-type drop method to drop MAI solution can reduce the damage of MAI solution to PbI 2 thin films, and high-concentration MAI solution can also transform PbI 2 film into high-class perovskite film via our route. Moreover, a perovskite solar cell was prepared via the modified sequential deposition route through LLLD under absolute ambient conditions and got a 10.7% PCE in a size of 0.2 cm 2 , which is the highest efficiency ever recorded in perovskite solar cells with such a spongy carbon/FTO composite counter electrode. Obviously, a great deal of production of high-efficiency perovskite solar cells, with a low-cost spongy carbon/FTO composite counter electrode, under an absolutely ambient atmosphere, is possible.

Conflicts of Interest:
The authors declare no conflict of interest.