Hybrid organic–inorganic perovskite, the chemical general formula ABX₃ (A = CH₃NH₃, HC(NH₂)₂, Cs; B = Pb, Sn; X = Cl, Br, I) perovskite family, is a well-known material with excellent absorption coefficient,^[1,2] long carrier diffusion length,^[3–5] high mobility,^[6,7] and a tunable bandgap.^[8–10] These properties can be tuned by changing the A, B, and X atoms. Nowadays, the certified efficiency of a hybrid organic–inorganic perovskite solar cell (PSC) has already reached up to 22.1%,^[11] which proves that the hybrid organic–inorganic perovskite is a viable candidate for crystalline silicon solar cells and copper indium gallium selenide (CIGS) commercial photovoltaic (PV) devices, owing to its low-cost and facile solution-based fabrication process.

The perovskite film is at the core of PSC, whose quality greatly determines the photovoltaic performance. Many photo-physical properties, such as light harvesting, charge carrier transport and diffusion length, can be dramatically affected by the crystallization of the perovskite.^[12–14] The defects and the crystal grain boundaries of the perovskite crystallites act as the traps of carriers, which aggravate the charge recombination. The current density–voltage (J–V) hysteresis effect, which causes inaccuracy in evaluating cell efficiency, is also believed to be related to the crystallinity and interfaces of the perovskite.^[15,16] Thus, the performance of perovskite solar cell strongly depends on the morphology and composition.^[17–22] So far, many methods have been exploited to prepare perovskite films with excellent photon and electron properties through tuning nucleation growth and the resulting micro-morphology of perovskite films.^[23] Compared with the one-step method, the two-step method offers the good controlling of morphology, and it is much more economical and convenient than the vapor deposition method. For the two-step method, an inorganic halide (e.g., lead (II) iodide, PbI₂) layer is first deposited, whereupon organic ions (e.g., methyl ammonium iodide, MAI) are intercalated between the layers of edge sharing PbI₆ octahedra and form the perovskite. Burschka et al. used this two-step method to fabricate a perovskite solar cell in 2013,^[24] wherein the spin-coated and dried PbI₂ layer was dipped in an appropriate 2-propanol solution with MAI. The process for PbI₂ deposition in the sequential deposition was found to be crucial to the final MAPbI₃ morphology and photovoltaic performance.^[25] As a perovskite inorganic framework, the morphology of the original PbI₂ may affect the reproducibility of the photovoltaic performance because high-quality perovskite thin film depends on the morphology of the PbI₂ layer, dipping time, concentration and temperature of the MAI solution. In addition, the MAI penetration is a serious issue due to the volume expansion when the MAI is inserted into the PbI₂ crystal lattice.^[25,26] This will lead to incomplete transformation of PbI₂ to perovskite on account of the fact that the initially formed MAPbI₃ compact layer blocks the MAI from being diffused and inserted into the PbI₂ layer. Moreover, the residual PbI₂ has detrimental effects on the carrier separation and transportation. In the present work, we fabricate the materials with different surface morphologies (mainly surface coverage and different roughness) by adjusting the drying process of the PbI₂ layer.^[26] Then the influences of synthesis conditions on the microstructures of perovskite films and the optical-electrical performances of the cells are investigated.

Patterned fluorine-doped tin oxide (FTO) substrates were cleaned in the solvents of detergent solution, acetone, ethanol and deionized water sequentially. Then the cleaned substrates were treated by UV-O₃ for 15 min to make them well hydrophilic. After that, 50-nm thick compact TiO₂ (c-TiO₂) was formed on the cleaned FTO glass by spin-coating 0.15-M solution of titanium diisopropoxide bis (acetylacetonate) (75 wt% in isopropanol, Aldrich) in 1-butanol (99.8%, Aldrich) at 3000 r.p.m for 30 s, which was dried at 125 °C for 5 min, then the process was repeated once again. A mesoporous TiO₂ (m-TiO₂) layer composed of 20 nm-sized particles was deposited by spin-coating at 3000 r.p.m for 30 s by using a commercial TiO₂ paste (18-NRT, Dyesol) diluted in ethanol (2:7, weight ratio). After drying at 125 °C for 10 min, the substrate was annealed at 500 °C for 30 min.

2.1. PbI₂ layer fabrication

PbI₂ (99.99%, Sigma-Aldrich) was dissolved in several kinds of polar solvents (N, N-dimethylformamide, DMF, , Sigma-Aldrich; N, N-dimethylsulfoxide, DMSO, 99.5%, Sigma-Aldrich) under stirring at 70 °C until its complete dissolution. The PbI₂ precursor solution was spin-coated on the substrates (FTO/c- TiO₂/m-TiO₂) at 3000 r.p.m for 30 s within an N₂ filled glovebox (deposition spinning details can be seen in Fig. 1). Also, different solutions used were simplified with A, B, C, and D, respectively, and then PbI₂ films with different solutions would be prepared after the coating process. The precursor solution of 1-M PbI₂ in DMSO and DMF/DMSO (92/8 v/v) mixture, which are marked with PbI₂-DMSO and PbI₂-DMF & DMSO, respectively. The preparation details of the PbI₂ precursor film with 1-M final concentration are shown in Table 1.

Fig. 1.

	Figure Option ViewDownloadNew Window
	Fig. 1. (color online) A schematic diagram representation of the fabrication process.

Table 1.

Table 1.

Table 1. The PbI2 precursor films prepared with different processes. . Method Precursor solution PbI2/g PbI2 film formation process I A DMF & DMSO (92:8, v/v) 0.462 PbI2 solution was deposited by spin coating at 3000 r.p.m for 30 s under an N2 atmosphere within a glovebox, followed by immediately heating on a hot plate at 80 °C for 15 min. B DMF & DMSO (92:8, v/v) 0.462 After the solution of PbI2 was spin-coated, the wet film was kept in a covered Petri dish at room-temperature within a glovebox for 10 min and then heated on a hot plate at 80 °C for 5 min. II C DMSO 0.462 Before a spin coating procedure the substrates were preheated at 80 °C, and the PbI2 solution was deposited by repeating step a. D DMSO 0.462 Repeating step a.

Table 1. The PbI2 precursor films prepared with different processes. .

The PbI₂ morphology-related reaction kinetics in the two-step method is assigned to heterogeneous nucleation, which leads to the formation of PbI₂ nuclei on the m-TiO₂ stack layers. The liquid nature of the film, augmented by the presence of a large amount of solvent, leads to subsequent crystal growth, where the extent of growth is controlled by adjusting the nucleation density and growth condition. Figure 2 shows the top views of PbI₂ precursor films processed under different deposition conditions, each of which is composed of platelet-shaped PbI₂ crystals that are between 100 nm and 400 nm in diameter. It is clear that there is a big difference in surface morphology, which is due to their different fabrication processes. As seen clearly from Figs. 2(a) and 2(b), the slow crystallization process prefers to generate polyporous crystals because of the slow solvent evaporation. The drying process of PbI₂ film also helps to create larger pores and voids in the PbI₂ crystals which penetrate through the whole film. This porous morphology is probably related to the controlled slow solvent evaporation in the sealed box. These loose cavities can enhance the penetration of MAI in the PbI₂ layer during dipping, leading to the formation of perovskite. Thus, it can be seen that the morphology of the PbI₂ film is likely to be influenced by the low-temperature drying process. Compared with the nanoparticles in mixed solvent, the nanoparticles in pure DMSO solvent have sizes of 250 nm–400 nm and stack with a PbI₂ layer on the m-TiO₂ film as shown in Figs. 2(c) and 2(d). The majority of voids are formed within the PbI₂ layer, which is probably attributed to the high viscosity DMSO solvent and low nucleation density during the spin-coating process. Especially, the particulate network of PbI₂ is very much sparser in appearance (Fig. 2(d)), and the surface coverage on the TiO₂ film is smaller than that in the case of PbI₂ formed by using the substrate-preheating technique. The coverage seems to be related to competition between the solvent evaporation rate and the PbI₂ nucleation rate, that is, the PbI₂ nucleation rate becomes higher than the solvent evaporation rate because the substrate pre-heating process increases the nucleation sites. In the case of PbI₂ fabricated through process D, too much of a large gap forms in the grain and it is speculated that the MAPbI₃ films could not cover the substrates completely during the following deposition process, which will be discussed later in this work.

Fig. 2.

	Figure Option ViewDownloadNew Window
	Fig. 2. SEM images of PbI2 fabricated with method A, B, C, and D, respectively.

The DMSO acts as both a solvent and a coordination reagent in the formation of a complex of PbI₂·DMSO, while DMF behaves only as a solvent with a higher evaporation rate than DMSO (vapor pressure data are listed in Table 2). Because of the relatively low evaporation rate and high viscosity of pure DMSO, the coverage of the substrates is very poor, which is unfavorable for the formation of perovskite in the second dipping step. Meanwhile, the coverage of PbI₂ film decreases if no pre-heating process is employed, which might be attributed to the hydrophobicity and large viscosity of DMSO. For mixed solvents, DMF solvent possesses a lower vapor pressure at room temperature which affects the morphology of the obtained PbI₂ film via the slow crystallization process.

Table 2.

Table 2.

Table 2. Typical physical parameters for DMF and DMSO solvents used for PbI2 precursor. . Solvent Density/(g/mL) Boiling Point/°C Viscosity/mPa·s Vapor pressure/(mmHg, 20 °C) DMF 0.948 152–154 0.92 2.7 DMSO 1.100 189 2.0 0.42

Table 2. Typical physical parameters for DMF and DMSO solvents used for PbI2 precursor. .

A series of MAPbI₃ films is prepared from different PbI₂ precursor films. For comparison, these perovskite thin films are denoted as PSC-a, PSC-b, PSC-c, PSC-d, respectively. Figure 3 shows the SEM images of the MAPbI₃ films prepared. It can be seen that the grain sizes and coverage rates of the resulting perovskite films change significantly. All the samples exhibit similar cuboid crystals. However, PSC-a and PSC-b overlayer have better quality than PSC-c and PSC-d in the aspect of smoothness and compactness, especially PSC-b is nearly pinhole-free and has a relatively uniform surface (size distribution is shown in Fig. 3(f)). The PSC-a appears to be composed of small sized perovskite crystals ( ), and its surface is covered with few uneven large cubic grain. The formation of coarse perovskite crystal usually appears during the Ostwald ripening process because of the dissolution-re-crystallization process (i.e., smaller perovskite crystals dissolve and re-deposit onto larger perovskite crystals). Obviously, the existence of slow crystallization at room temperature, with the same dipping time, is favorable for the large nanoparticles (cubic shape) on the surface MAPbI₃ film, an observation that is assistant and supports that the dense PbI₂ morphology impedes the penetration of MAI, so a portion of the formed small perovskites nanoparticles most likely continue to evolve by the above proposed mechanism.

Fig. 3.

Figure Option ViewDownloadNew Window

Fig. 3. SEM images of MAPbI3 fabricated with different methods. ((a) and (b)) MAPbI3 fabricated with method A; (c) the size analysis of the sample shown in panel (b); ((d) and (e)) MAPbI3 fabricated with method B; (f) size analysis of the sample shown in panel (e); ((g) and (h)) MAPbI3 fabricated with method C; (i) size analysis of the sample shown in panel (h); ((j) and (k)) MAPbI3 fabricated with method D; (l) size analysis of the sample shown in panel k.

In brief, the drying process at room temperature enhances the formation of larger pores in the PbI₂ film and larger MAPbI₃ particles, indicating the complete transformation of PbI₂ into MAPbI₃. In the case of pure DMSO used (PSC-c and PSC-d), the MAPbI₃ films are mainly composed of large crystals ( ) filled with -nm small crystals, indicating the uneven coverage of the perovskite film. Without the preheating deposition, the gaps in perovskite film are formed (Fig. 3(k)) because localized nucleation gives rise to leakage current and direct contact between electron-and hole-transporting layers, resulting in serious charge recombination and deterioration of light absorption. These results reveal that the preheating of substrate has an important effect on PbI₂ spin coating through ameliorating the surface morphology and improving the quality of the resulting MAPbI₃ films.

Taking it into consideration that the reaction between PbI₂ and MAI is a solid–liquid reaction, we believe that the reaction rate is determined by three major factors: (i) property of PbI₂ film, such as preferential orientations of crystal, crystallinity and morphology of PbI₂ precursor film; (ii) concentration of MAI/IPA solution; (iii) temperature of the reaction system. In our study, the latter two factors have been carefully controlled. So we believe that the great difference among PbI₂ films in the preparation of perovskite comes from the morphology difference as shown in Fig. 2. Thus, controlling the morphology of PbI₂ may adjust the number of nuclei of the perovskite, and then affect the morphology of perovskite and grain size. We control the morphology of PbI₂ by adjusting the solvent polarity and drying process. Mainly, two main factors contribute to the faster conversion of nanoporous PbI₂ into perovskite rather than compact PbI₂: firstly, the interpenetrating nanoporous morphology enables the occurrence of reaction in the entire film simultaneously; secondly, the crystallite size is lower in the nanoporous PbI₂ than those in the compact PbI₂, shortening the diffusion length for MAI. In a word, the reaction is limited by the interface of PbI₂(s)–MAI (l) and fresh MAPbI₃, so the morphology of PbI₂ film is affected indeed in the two-step method because a large volume expansion occurs when PbI₂ converts into MAPbI₃ (about 74.8 cm³–149.0 cm³) due to the density difference between originally edge-sharing octahedral PbI₂ (6.16 g/cm³) and the corner-sharing octahedral perovskite (4.29 g/cm³).

In general, GIXRD is an effective tool to determine the surface crystallinity by changing the incident angle of x-ray. The samples are kept in a carefully controlled measurement condition, so the diffraction peak intensity is mainly determined by the crystallization, densification and components of the sample. The GIXRD patterns of the MAPbI₃ films obtained from PbI₂ are shown in Fig. 4. All film samples consisting of a cubic crystal phase possess similar patterns but with different diffraction intensities. The strong diffraction peaks at 14.21°, 20.68°, 24.57°, 28.52°, 31.90°, 35.04°, 40.62°, and 43.19° could be ascribed to (002), (022), (222), (004), (042), (422), (044), and (006) planes of the MAPbI₃ perovskite, respectively. Apart from the cubic character of an obvious perovskite phase with the space group of F23, a small number of PbI₂ phases exist in the PSC-d sample (the rose red curve in Fig. 4), and the peak at 12.73° corresponds to the (001) lattice plane of PbI₂, which indicates incomplete conversion of PbI₂. In combination with the SEM result, we speculate that the rough and nonuniform stacking nature of PbI₂ crystal retards the diffusion of MAI and therefore results in the subsequent incomplete conversion of PbI₂.

Fig. 4.

	Figure Option ViewDownloadNew Window
	Fig. 4. (color online) XRD patterns of MAPbI3 fabricated with methods A, B, C, and D, respectively.

The higher intensity in the PSC-b film should be attributed to the slight thickness of the sample, or uniformly distributed nanocrystals and complete crystallization of the sample. Furthermore, the slight differences in lattice constant and full widths at half maximum (FWHMs) for different films imply that the PbI₂ precursor film only influences the perovskite film coverage and morphology without causing significant changes in the crystalline structure of the material.

Table 3 shows the lattice constants and FWHMs of the diffraction peak at . The average grain size D of the MAPbI₃ nanocrystals can be estimated from the (002) peak at 14.2° by using the Scherrer equation: , where k is the Scherrer constant (equal to 0.89), λ is the x-ray wavelength (equal to 0.154 nm), θ is the Bragg diffraction angle, and β is the FWHM of the (002) diffraction peak. The FWHMs of the (002) diffraction peaks are 0.350, 0.381, 0.423, and 0.377, which correspond to average grain sizes of 22.6, 20.8, 18.7, and 21.0 nm for the PSC-a, PSC-b, PSC-c, and PSC-d sample, respectively.

Table 3.

Table 3.

Table 3. Lattice constants of MAPbI3 films. . Sample a/Å Vol/Å3 FWHM Grain size PSC-a 12.560 1981.385 0.350 22.6 PSC-b 12.578 1989.916 0.381 20.8 PSC-c 12.618 2008.961 0.423 18.7 PSC-d 12.582 1991.815 0.377 21.0

Table 3. Lattice constants of MAPbI3 films. .

Figure 5 shows the absorption spectra and PL spectra of the corresponding PSC films. Their absorptions cover a wide range of light wavelengths from visible to the near-infrared region. All of these films exhibit the same absorption shoulders near 750 nm, which is typical nature for the MAPbI₃ phase. As a result of the poor coverage of the substrate by film when MAPbI₃ is prepared with pure DMSO solvent, their absorptions are lower than those of the MAPbI₃ fabricated with mixed solvents, which is more prominent in the short wavelength region (350 nm–550 nm). The PSC-b shows the most enhanced absorption, which is related to the improved morphology and coverage rate. In addition, it can be seen that MAPbI₃ films prepared with DMF/DMSO mixed polar solvent have strong absorptions, indicating that more PbI₂ nanoparticles are converted into MAPbI₃ nanoparticles. The increase in absorption supports that the loose structure assembled from PbI₂ nanoparticles allows MAI to reach the interior of PbI₂ very easily and facilitates the complete transformation of PbI₂. The characteristic absorption shoulder near 750 nm for PSC-c film increases slightly, which reflects the stronger light-scattering effect associated with abnormally large grains. This result reveals that good surface morphology of the MAPbI₃ film is important for the light absorption of the perovskite solar cell. Figure 5(b) shows the steady-state PL spectra of perovskite films excited at 532 nm. The PL quenches at around 776 nm, which is consistent with the behavior of the optical gap from the absorption spectra. These results suggest that the controlled growth of PbI₂ scaffolds could be a useful new route to manipulating the morphology of the perovskite active layer, opening new possibilities for enhancing device performance.

Fig. 5.

	Figure Option ViewDownloadNew Window
	Fig. 5. (color online) (a) Absorption spectra and (b) PL spectra of MAPbI3 films.

The photovoltaic performances of the fabricated PSCs based on FTO/c-TiO₂/m-TiO₂/MAPbI₃/HTM/Ag architecture are examined and the current–voltage (J–V) characteristics are measured in the reverse bias condition ( ) under standard AM 1.5G illumination, with an effective area of 0.09 cm². All the performance parameters such as short-circuit current density ( ), open circuit voltage ( ), fill factor (FF) and power conversion efficiency (PCE) are summarized in Table 4, which are achieved by a reverse bias scan. The solar cell with the PSC-b perovskite film exhibits a highest PCE performance of 11.59% with of 21.86 mA/cm², of 0.882 V and of 60.73 as expected. Meanwhile, the PCE of PCS-b keeps a high level of more than 93% of the initial efficiency (from 11.59% to 10.87%) after 43-day storage in air (Fig. 6(b)).

Fig. 6.

	Figure Option ViewDownloadNew Window
	Fig. 6. (color online) (a) J–V curves of devices fabricated with different methods. (b) J–V curves of the best performance device fabricated by method B.

Table 4.

Table 4.

Table 4. Photovoltaic parameters of devices with different fabrication methods. . Fabrication methods /V FF /% A 0.860 18.24 57.56 9.03 B 0.882 21.86 60.07 11.59 C 0.757 15.51 44.44 5.22 D 0.993 10.92 36.28 3.92

Table 4. Photovoltaic parameters of devices with different fabrication methods. .

For mixed solvents, the higher PCE of PSC of the material prepared by method B is attributed to the PbI₂ scaffold with mesoporous structure, which promotes the conversion process of and less damage to perovskite. In addition, many studies have shown that a larger perovskite grain size can reduce the interfacial area, which can hinder charges from being collected. Larger grains also possess less bulk defects and higher mobilities, so the photo-generated carriers could be collected more easily.^[27,28] We note that Yi et al.^[29] reported solar cells based on bilayer porous lead halide (PbX₂, containing PbI₂ and PbBr₂) films prepared by using PbX₂ solutions of DMF/DMSO mixed-solvent. The MAI solution can easily diffuse through the PbI₂ film and react with PbI₂ more completely because of the continuous mesoporous structure. At the same time, these pores also provide the space needed to accommodate the large volume expansion and the insertion of the organic cations into the PbI₂ lattice. The interpenetrating nanoporous morphology enables fast penetration of the MAI solution during the conversion step, and perovskite grains can grow freely and form a compact MAPbI₃ film. For pure DMSO solvent, PCE of 3.92% is achieved without pre-heating and increases to 5.22% when the substrate is heated to 80 °C. The pre-heating treatment can lead to the increase of photocurrent and voltage. By contrast, the PbI₂-DMSO based device suffers relatively low and FF, which is attributed to the poor coverage with high leak current and low light-harvesting. The coverage of PbI₂ film on the m-TiO₂ layer decreases when only DMSO is used as solvent, because the hydrophobicity and high viscosity of DMSO make it difficult to spread on the substrate surface. In addition, since the substrate surface can provide nucleation sites for the nucleation of crystals, pre-heated substrate is likely to play a positive role in nucleating. And a higher coverage of perovskite layer is beneficial for the light absorption, leading to higher photocurrent and voltage.

In this work, a suitable polarity solvent, processed with and without drying at room temperature, pre-heating process of substrate for PbI₂ deposition in the two-step coating procedure, and the affinity to PbI₂, can facilitate the formation of mesoporous PbI₂ films and the fabrication of a uniform and crystalline perovskite film. In the case of full coverage and less pinholes of the perovskite film, excellent device efficiency and stability are achieved successfully. Meanwhile, the PCE of PCS-b keeps a high level of more than 93% (from 11.59% to 10.87%) after the cell has been used for 43 days under ambient conditions.

The authors thank the postdoctoral research station of the Hefei Institute of Physical Science for their technical support.