Rapid crystallization and controllable growth of perovskite thin films via a seeded approach

Long diffusion length, suitable bandgap, and high absorption coefficient of organic–inorganic halide perovskites have attracted great attention in developing perovskite solar cells (PVSCs).^1–9 To fabricate cost-effective and efficient PVSCs, rapid crystallization with controlled morphology of perovskite films is the key.^10–12 Usually, lengthy thermal annealing is used for crystallization of lead iodide (PbI₂) and methylammonium iodide (CH₃NH₃I, MAI) precursors into perovskite crystals (CH₃NH₃PbI₃, MAPbI₃), In this study, we report a rapid crystallization technique through seeded approach for perovskite formation with controlled morphology and grain size. A perovskite seeding layer was formed by incubating a low concentration of MAI solution on top of the PbI₂ thin film prior to introducing a high concentration of MAI solution. The perovskite seeding layer is able to accelerate the nucleation and growth of perovskite to generate dense and uniform films with tunable crystal grain size and morphology.

PbI₂ (beeds, 99.9999%) was purchased from Sigma-Aldrich. MAI was purchased from Dyesol. 2-propanol alcohol (IPA) and dimethylformamide (DMF) were purchased from VWR. Poly(3,4-ethylenedioxythiophene):poly(styrenesulphonate) (PEDOT:PSS) and Phenyl-C61-butyric acid methyl ester (PCBM) were purchased from Alfa Aesar. All the chemicals were used as-received without any further purification.

PbI₂ was dissolved in DMF at a concentration of 1 mol/l. MAI was dissolved in IPA with different concentrations from 0.5 to 30 mg/ml. Both solutions were heated at 70 °C overnight to make sure the complete dissolution of MAI and PbI₂, respectively. The PbI₂ solution was spun on the PEDOT:PSS substrate at 80 °C at 4000 rounds per second (rpm) for 30 s. The low-concentration MAI solution (0.5, 1, 1.5, or 2.0 mg/ml) was dropped onto the surface of PbI₂ for 1 min, and the remaining solution was spun off at 4000 rpm for 75 s. The high-concentration MAI solution (10, 20, or 30 mg/ml) was subsequently dropped onto the PbI₂/seed film for 1 min, and the remaining solution was spun off at 4000 rpm for 75 s.

Substrate cleaning was done by cleaning and rinsing in detergent followed by sonication in deionized water, acetone, and isopropanol for 40 min sequentially. Substrates were then subjected to oxygen plasma for 30 min, followed by being heated on a hot plate at 80 °C for 5 min. PEDOT:PSS films were spin coated at 5000 rpm for 45 s and annealed at 140 °C for 15 min. Before spin-coating the PbI₂ layer, both substrate and PbI₂ solution were maintained at 80 °C. A 1M (precisely 461 mg/ml) solution of PbI₂ dissolved in anhydrous dimethylformamide was spin coated on PEDOT:PSS films at 3000 rpm for 10 s and 6000 rpm for 10 s and subsequently annealed at 100 °C for 10 min. For the formation of seeded layer, 1 mg/ml low-concentration MAI in IPA was drop casted on PbI₂ films and left undisturbed for 1 min before spinning off the remaining solution at 4000 rpm for 75 s. A high-concentration (10 mg/ml) MAI solution was then drop casted on the seeded layer and remained for 1 min followed by spinning off left solution at 4000 rpm for 75 s and then followed by natural dry for 30 min. The reference perovskite solar cells have perovskite thin films thermally annealed on a hot plate at 100 °C for 10 min after spin-coating at 4000 rpm for 45 s. A 20 mg/ml of PCBM solution was then spin coated at 2000 rpm for 30 s and annealed at 100 °C for 15 min. The devices were completed by thermal evaporation of 100 nm Al at 2 Å/s.

X-ray diffraction (XRD) measurements were performed with a Bruker D8 X-ray diffractometer with a conventional cobalt target x-ray tube set to 40 kV and 30 mA. The single path absorption was measured using a Beckman D800 UV-Visible spectrometer. Scanning electron microscopy (SEM) images were taken by using a Joel 7000 scanning electron microscope.

For a typical seeded deposition of perovskite, PbI₂ and MAI were dissolved in DMF and 2-propanol (IPA), respectively. 100 μl of 1M PbI₂ was spun onto PEDOT:PSS covered indium tin oxide (ITO) glass, followed by dropping 200 μl of low-concentration MAI (1 mg/ml) on the surface of PbI₂ thin film and incubating for 1 min to form the perovskite seeds. The remaining MAI solution was spun off and followed by dropping a high concentration of MAI solution (10 mg/ml) on the seed layer and incubating for 1 min before spin-coating. The yellowish color of PbI₂ did not clearly change after the addition of a low-concentration MAI solution. However, a dark brown color appeared immediately after the introduction of high-concentration MAI solution, indicating the formation of perovskite crystals.

Figure 1 shows the topography images recorded by SEM for the PbI₂ thin film [Fig. 1(a)], PbI₂ thin film with perovskite seeds [Fig. 1(b)], and perovskite thin film obtained by seeded growth [Figs. 1(c) and 1(d)]. SEM image in Fig. 1(a) reveals that the PbI₂ film has a continuous and/or porous morphology, while the PbI₂/seed film [Fig. 1(b)] is purely porous due to the reaction between PbI₂ and low-concentration MAI, and the subsequent formation of MAPbI₃ seeds. After the addition of high-concentration MAI, a compact and pinhole-free perovskite thin film is formed, as shown in Figs. 1(c) and 1(d), respectively.

To further elucidate the seeded growth behavior of perovskite thin films, we have characterized the phase structure of films using XRD, as shown in Fig. 2(a). The XRD pattern of PbI₂ film shows a strong peak at 14.6°. After the addition of low-concentration MAI solution, the resultant thin film generates an XRD pattern basically the same as the pattern of pure PbI₂, which may indicate that the perovskite seeds are amorphous and/or nanocrystalline in nature. In the XRD pattern of perovskite thin films formed by seeded growth, the peak at 14.6° completely disappears, suggesting the complete conversion of PbI₂ into perovskite. There is also no peak shown at 12°, the characteristic peak of MAI, indicating that there is no residue of MAI on the perovskite films grown from seeded technique. More importantly, these unannealed thin films from seeded growth show similar XRD patterns as those annealed at 100 °C for 60 min. It is anticipated that the incubation of high-concentration MAI solution on the top of the porous PbI₂ film with MAPbI₃ seeds facilitates the infiltration and diffusion of MAI and thus promotes the nucleation and growth of perovskite thin films. Overall, the seeded approach facilitates the nucleation and growth of perovskite films without requiring lengthy thermal annealing for crystallization.

The optical property of the seeded growth of perovskite thin films has also been characterized using UV-vis measurements, as shown in Fig. 2(b). Both PbI₂ and PbI₂/seed thin films show strong peaks at ∼496 nm, which can be attributed to the strong absorption of PbI₂. Both PbI₂ and PbI₂ with perovskite seed films also show similar absorption edges, although there is a difference in absorption intensity, which may be derived from the varied topography as shown in Figs. 1(a) and 1(b), where a porous film was obtained in the PbI₂ film with perovskite seeds. For perovskite thin films with or without thermal annealing obtained by the seeded growth, UV-Vis spectra show an absorption edge at 780 nm corresponding to the bandgap of perovskite film at ∼1.60 eV. The annealed perovskite thin film presents an absorbance spectrum like that of the unannealed thin film, further proves that the seeded approach could achieve rapid crystallization and produce perovskite thin films without the necessity of thermal annealing.

To optimize the seeded crystallization, we have extensively tailored the MAI concentration and incubation time during the synthesis of the perovskite films. To begin with, we have investigated the influence of the first step low-concentration MAI solution on the generation of perovskite seeds by varying its density from 0.5, 1, 1.5, to 2 mg/ml while fixing the density of the second step high-concentration MAI (e.g., 10 mg/ml), as shown in Fig. S1 in the supplementary material.²¹ When the density of low-concentration MAI is 0.5 mg/ml, besides the formation of a dense and pinhole-free perovskite thin film, a large number of small particles also exist. When the low concentration of MAI is 1.5 mg/ml, it generates some large clusters. Further increasing the MAI concentration to 2 mg/ml results in the formation of large perovskite particles, instead of amorphous/nanocrystalline perovskite seeds. Thus, the desired density of low-concentration MAI is determined as 1 mg/ml, which gives the perovskite thin films with the most uniform morphology. Moreover, the density of the second step high-concentration MAI was also optimized. The morphologies of perovskite films obtained by adjusting the concentration of second step high-concentration MAI solution from 10, 20, to 30 mg/ml while keeping the density of low-concentration MAI solution constant at 1 mg/ml are shown in Fig. S2 of the supplementary material.²¹ As observed, perovskite thin films with the best morphology were fabricated at a concentration of MAI 10 mg/ml, while a higher density of MAI causes generation of rough perovskite surface with big perovskite particles. Through this optimization, we concluded that 1 mg/ml of the first step low-concentration MAI to form the perovskite seeds and 10 mg/ml of following high-concentration MAI would produce uniform, dense, and pinhole-free perovskite thin films without unwanted particles or clusters on the surface.

Furthermore, we have found that the crystal domain size of perovskite thin films can be well manipulated by adjusting the incubation time of low-concentration MAI solution on the top of PbI₂ thin film, while keeping the deposit parameters for the high-concentration MAI constant. Based on the XRD patterns of perovskite thin films shown in Fig. 3(a), we have analyzed the crystal domain size by using the Scherrer equation.¹³ The crystal domain size of perovskite thin film increases from 11.7, 15.6, 16.9, 17.5, to 20.9 nm when the incubation time changes from 10, 20, 30, 40, to 60 s, respectively. Further regression analysis shows that the crystal domain size is linear to the incubation time of low-concentration MAI solution on the surface of PbI₂ thin film, as shown in Fig. 3(b). This is an encouraging finding, as it gives a clue to fine tune the crystal domain size of perovskite thin film by controlling the incubation time of low-concentration MAI on PbI₂ film. The increase of crystal domain size with the elongation of incubation time is in agreement with surface morphology observed from SEM images. As shown in Fig. S3 of the supplementary material,²¹ the grain size of perovskite films increases with gradually increasing the incubation time from 10 to 60 s. We have randomly selected 100 grains in each SEM image to analyze the size distribution. The size of perovskite grains is 261 ± 62 nm, 311 ± 58 nm, 349 ± 63 nm, and 835 ± 260 nm, corresponding to an incubation time of 10, 30, 40, and 60 s, respectively, as shown in Fig. 3(c). Nevertheless, the incubation time is not allowed to be too long (e.g., >1 min). For instance, when the incubation time is 90 or 120 s, the PbI₂ could not fully convert into perovskite, evidenced by the remaining PbI₂ peak at 14.6° shown in Fig. 3(a). To elucidate the potential reason, we have carried out another series of XRD measurements to investigate the interaction between PbI₂ and the low-concentration MAI, without introducing the second step high-concentration MAI. As shown in Fig. S4 of the supplementary material,²¹ for an incubation time of less than 1 min, XRD patterns show no clear difference between PbI₂ with perovskite seeds and pure PbI₂. However, for a longer time of incubation (90 and 120 s, respectively), the perovskite peak intensity at 2 theta 39.5° becomes increasingly stronger, suggesting increased crystallinity of perovskite seeds along with increased incubation time. It is anticipated that perovskite seeds with enhanced crystallinity could hinder the reaction of underlying PbI₂ with high-density MAI, resulting in PbI₂ residues, while amorphous/nanocrystalline perovskite seeds from incubation time less than 1 min facilitate the complete conversion of PbI₂ with high-density MAI into perovskite.

Note that the seeded perovskite growth presented here is different from the two-step interdiffusion method.¹⁰ In the interdiffusion method, the perovskite thin film is formed by a solid-state reaction of MAI thin film and the underneath PbI₂ at a temperature of 100 °C for a certain period of time. To obtain a pure perovskite thin film, a thickness ratio of 1.4:1 for the MAI layer:PbI₂ layer is needed to form a stoichiometry iodine perovskite. Also, thermal annealing is required to promote the interdiffusion of solid-state MAI and PbI₂ to form perovskite, as evidenced by the gradual change of color from light brown to dark brown along with thermal annealing. In this seeded growth approach, the addition of MAI was divided into two steps. In the first step, the low-concentration MAI solution reacts with PbI₂ to form a seeded layer of MAPbI₃, which grows into a perovskite thin film upon the addition of high-concentration MAI solution. Without the first step low-concentration MAI, the high-concentration MAI could react with PbI₂ immediately such that a dense perovskite thin film will be formed upon the addition of MAI which hinders the diffusion of MAI into PbI₂ and causes the incompletion of PbI₂ conversion. Thus, thermal treatment is needed to further drive the diffusion of MAI into deep PbI₂ layer to achieve desired perovskite films. In this work, with the help of low-concentration MAI solution to form the initial seeding perovskites, the compact PbI₂ thin film rearranges and becomes more porous, evidenced by SEM images in Fig. 1, which allows the easy diffusion of the following high-concentration MAI solution into the PbI₂ network, and the perovskite grain growth barrier energy can be greatly reduced, resulting in rapid crystallization of perovskite without resorting to lengthy thermal annealing.

After optimization of perovskite active layer through the seeded approach, PVSCs with device architecture of ITO glass/PEDOT/Seeded Perovskite/PCBM/Al were fabricated. The electrical performance was evaluated by means of current density versus voltage (J-V) measurements under 1.5 G (100 mW cm⁻²) irradiation. Figure 4(a) shows a typical J-V characteristic from PVSCs with natural dry of perovskite layer for 30 min. It has an open-circuit voltage (V_OC) of 0.8 V, a short-circuit current density (J_SC) of 23.9 mA cm⁻², a fill factor (FF) of 55.2%, and an overall power-conversion efficiency (PCE) of 10.5% in the reverse sweep direction, and a V_OC of 0.8 V, a J_SC of 22.3 mA cm⁻², an FF of 51.3%, and an overall PCE of 9.1% in forward scan. The reference devices, which were thermally annealed on a hot plate at 100 °C for 15 min, show better performance [Fig. 4(b)] with PCE of 12.9% in reverse scan and around 9.1% in forward scan, respectively. The improved performance from thermal annealing could be attributed to completely drying out of an organic solvent in the perovskite layer. Overall, the perovskite solar cells obtained via the seeded growth exhibit decent efficiency as they bear the simplest device configuration without any additives in the perovskite active layer and any sort of interface engineering.^14–20 

In conclusion, we demonstrate a novel seeded approach for rapid crystallization of perovskite films without the need of lengthy thermal annealing. The perovskite seeds formed by placing a low-concentration MAI solution on the top of PbI₂ thin film regulate the growth of dense and uniform perovskite thin films with controllable grain size. PVSCs fabricated through this seeded approach exhibit a decent power-conversion efficiency. This facile seeded approach for the deposition of perovskite thin films with tunable grain size can facilitate the fabrication of cost-effective and efficient perovskite solar cells.

This work was partially supported by the National Science Foundation (NSF) (No. ECCS-1151140) and a postdoctoral research fellowship from the University of Alabama. The authors acknowledge the generous help from Zhiqun Lin, Xueqin Liu, and Xun Cui at the Department of Materials Science and Engineering, Georgia Institute of Technology. They thank the instrumental assistance from Central Analytical Facility (CAF) at the University of Alabama.