An efficient, flexible perovskite solar module exceeding 8% prepared with an ultrafast PbI2 deposition rate

Large-area, pinhole-free CH3NH3PbI3 perovskite thin films were successfully fabricated on 5 cm × 5 cm flexible indium tin oxide coated polyethylene naphthalate (ITO-PEN) substrates through a sequential evaporation/spin-coating deposition method in this research. The influence of the rate-controlled evaporation of PbI2 films on the quality of the perovskite layer and the final performance of the planar-structured perovskite solar cells were investigated. An ultrafast evaporation rate of 20 Å s−1 was found to be most beneficial for the conversion of PbI2 to CH3NH3PbI3 perovskite. Based on this high-quality CH3NH3PbI3 film, a resultant flexible perovskite solar sub-module (active area of 16 cm2) with a power conversion efficiency of more than 8% and a 1.2 cm2 flexible perovskite solar cell with a power conversion efficiency of 12.7% were obtained.

ScienTific RepoRts | (2018) 8:442 | DOI: 10.1038/s41598-017-18970-y sequential deposition method with evaporated PbI 2 and a CH 3 NH 3 I solution 16 , yielding a semi-transparent solar cell with a steady-state efficiency of 14.2% along with a 72% average transmittance in the near-infrared region, which ensures that this strategy has a good prospect.
Herein, a hybrid sequential deposition method was adopted to fabricate module-sized perovskite films on flexible substrates. The effects of the PbI 2 deposition rates on the quality of the perovskite films and device performance were investigated. At an ultrafast rate of 20 Å s −1 , large-area flexible perovskite solar cells with an active area of 1.2 cm 2 were prepared that achieved an efficiency of 12.7%. In addition, a flexible perovskite solar module with an active area of 16 cm 2 on a 25 cm 2 ITO-PEN substrate exhibited a PCE of more than 8%. Combining the merits of the low cost, short processing time, full coverage of the substrate, and high module performance, the PbI 2 film fabricated via ultrafast thermal evaporation holds promise for facilitating the development of industrial-scale perovskite solar cells.

Results
In this research, fullerene C 60 was selected as an electron transport material (ETL) and first evaporated onto a indium tin oxide-coated polyethylene naphthalate (ITO-PEN) flexible substrate 17 . Then, PbI 2 was evaporated on this amorphous C 60 layer at various controlled deposition rates within the range of 0.5 Å s −1 to 40 Å s −1 , which could be easily adjusted by controlling the evaporation source temperature. Using this thermal evaporation method, a uniform PbI 2 layer with a precisely controllable amount over a large area could be deposited.
Under the naked eyes, all these evaporated PbI 2 films exhibit a uniform color and an ultrasmooth surface (see Fig. S1 in the supplementary information). However, under scanning electron microscopy (SEM), the microscopic morphologies of the PbI 2 films vary slightly for the different evaporation rates. It has been reported that the structural properties of evaporated PbI 2 thin films can be adjusted by varying the deposition conditions, such as the substrate materials, substrate temperature and evaporation rates [18][19][20] . To impart more significant differences in the PbI 2 films and in the subsequent perovskite films and devices, the deposition rates were fixed at 0.5 Å s −1 (R 1 ), 20 Å s −1 (R 2 ) and 40 Å s −1 (R 3 ) for detailed investigation. Figure 1(a-c) shows the surface morphologies of the thermally evaporated PbI 2 films at deposition rates of R 1 , R 2 and R 3 . The PbI 2 film prepared with a deposition rate of 0.5 Å s −1 is smooth, compact and nearly free of pinholes, as shown in Fig. 1(a). In addition, at the rate of 20 Å s −1 , the PbI 2 film is still smooth but less compact. A porous surface and some pinholes can be seen in Fig. 1(b). When the deposition rate increased to 40 Å s −1 , the film became rough, and many PbI 2 nanoplates formed on the top surface, as shown in Fig. 1(c), which is similar to the morphology of the crystallized substrate 15 . These nanoplates are randomly arranged perpendicular to the substrate, which can create a large degree of porosity.
X-ray diffraction measurement (XRD) was employed to further investigate the microstructure of the PbI 2 films. As shown in Fig. 1(d), all the evaporated PbI 2 layers demonstrate a preferential (001) orientation compared with the raw PbI 2 powder (R 0 ). In the magnified XRD patterns, as the deposition rate rising up, the full width at half maximum (FWHM) of the peak increases slightly. To give a quantitative analysis, the average crystalline sizes of the PbI 2 films were calculated by the Scherer equation using the (001) diffraction peak through the formula reported elsewhere and shown in Supplementary Table S1. Clearly, these calculated values of crystalline sizes are smaller than the submicron-scaled grain sizes observed from the SEM images 19 . The PbI 2 film deposited at a rate of R 1 has the smallest (001) FWHM, meaning it has the largest crystalline size, which forms the most compact layer in Fig. 1(a). Compared with the R 1 -based sample, the R 2 -based sample has more pores in the thin film. However, upon further increasing the deposition rate to R 3 , a rough surface is formed. In brief, by varying the deposition rate, the morphology and microstructure of the PbI 2 film can be controlled in this work. Moreover, higher deposition rates correspond to shorter deposition times, which is good for manufacturing.
To understand the transformation from PbI 2 to MAPbI 3 perovskite, the deposition-rate-dependent PbI 2 films with different porosities were then transformed to perovskite films by spin-coating the MAI solution and annealing 21,22 . It has been demonstrated that perovskite films fabricated on an amorphous substrate, such as the [6,6]-phenyl-C61-butyric acid methyl ester (PCBM) layer, obtained by here described always contain residual PbI 2 16,23 . However, the performance of the final device can still be efficient if the amount of residue is controlled 14 . Figure 2(a-c) shows the XRD patterns of the different perovskite films. The PbI 2 peaks at 12.7° indicate the incomplete conversion from PbI 2 to MAPbI 3 , which causes variations in the stoichiometry of the resultant films. It was reported that the penetration depth of CH 3 NH 3 I into a compact PbI 2 layer is limited to a few tens of nanometer from the surface [24][25][26] . For the low deposition rate of R 1 , a distinct peak for PbI 2 can be seen in Fig. 2(a), which is ascribed to the inhibition of MAI diffusion caused by the formation of a compact MAPbI 3 perovskite film on the PbI 2 surface. At the rate of R 2 , the fabricated film shows a strong perovskite peak with the weakest peak for PbI 2 among the three samples. It is assumed that the pores on the surface of the PbI 2 film facilitate the permeation of MAI, which is beneficial for the conversion of PbI 2 to perovskite. Figure 2(e) shows that a uniform perovskite film without pinholes is formed across the entire SEM image of the top surface. In contrast, the R 3 -based film shows an obvious residual PbI 2 XRD peak that is even stronger than the perovskite peak, as illustrated in Fig. 2(c). Thus, a high-quality perovskite film could be fabricated by controlling the deposition rate of PbI 2 at R 2 .
As the key material in perovskite solar cells, the perovskite films fabricated with different PbI 2 deposition rates also have an important influence on the final photovoltaic performance. We first verified the principle of fabricating planar PSCs on a rigid substrate with the structure glass/ITO/C 60 /perovskite/spiro-OMeTAD/MoO 3 / Au 27 . The details of the fabrication process are described in the methods section. Except for the different deposition rates of PbI 2 films, the other conditions were kept the same for all cells. The photovoltaic performance of the PSCs based on different deposition rates of PbI 2 were measured under AM 1.5 G (100 mW cm −2 ) light illumination with a metal mask of 0.16 cm 2 , and illustrated in Fig. 3(a-d). The evolution of the device performance is consistent with the trends in the film and crystal quality observed from the SEM and XRD data. Compared with the devices fabricated at R 1 , all the photovoltaic parameters increased for the devices based on the R 2 film due to the improved conversion to perovskite. This result is consistent with research from other groups, where the optimized performance was obtained for a higher PbI 2 deposition rate 28 . However, when the deposition rate increased to 40 Å s −1 , the device PCE decreased due to the presence of more unreacted PbI 2 in the perovskite film. In contrast, the samples fabricated from the R 1 -and R 3 -based films exhibit smaller J sc values and lower FFs, With the high-quality perovskite thin film, small-area devices using the optimized method show an average PCE of 13.7% ± 0.7%. It has been reported that planar perovskite solar cells are more susceptible to the hysteresis effect 31,32 . Thus, the perovskite solar cells based on the R 2 deposition rate were measured under both reverse and forward scans and exhibited an almost negative hysteresis.
Considering that the materials and techniques used above are all processed at low temperature (not higher than 150 °C), we directly applied these techniques to an ITO-PEN substrate without any modification. The photovoltaic characteristics of the flexible MAPbI 3 devices with the above-optimized fabrication parameters and the same measurement conditions are shown in Fig. 4(a) and summarized in Table 1. The small-area flexible device shows PCEs of 13.9% under the reverse scan (V oc to J sc ) and 13.8% under the forward scan (J sc to V oc ). The inset of Fig. 4(a) shows a corresponding steady-state efficiency of 14.0% at a bias close to the initial maximum power point of 0.81 V. During a 60 second illumination period, the unencapsulated device shows a stable photocurrent density and PCE as a function of time. Figure 4(b) shows a typical external quantum efficiency curve (EQE) of the device. The integrated short-circuit current density is 18.5 mA cm −2 under the AM 1.5 G spectrum, which matches well with the J-V results. Upscaling the fabrication process, large area flexible device with an active area of 1.2 cm 2 can be obtained as in Fig. 4(d). It shows a V oc of 1.02 V, a J sc of 18.4 mA cm −2 , an FF of 0.68 and thus a PCE of 12.8% under the reverse scan. And under the forward scan, the device shows a V oc of 1.02 V, a J sc of 18.6 mA cm −2 , an FF of 0.66 and thus a PCE of 12.5%. The EQE spectra at three spots located at the center and two corners of the device were measured and show small variations in the J sc , as shown in Fig. 4(e), which provides evidence for the uniformity of the perovskite film over a large area. Compared with the small-area device, an increase in the active area from 0.16 cm 2 to 1.2 cm 2 resulted in a 4% lower FF, which we ascribe to the series resistance of the ITO. In addition, long term stability test of the flexible perovskite solar cells was performed and the cells showed promising result with only 10% PCE decrease over 1000 h in Fig. 4(c). Besides highly electron selectivity and low temperature process, C 60 as ETL has been proven with good stability in perovskite solar cells 17,33 .
From the results shown above, we can confirm that pinhole-free and uniform perovskite layers with good electrical properties were obtained using the sequential evaporation/spin-coating process, and therefore, flexible perovskite solar modules were fabricated. The module of the flexible perovskite solar cell had a large size of 5 cm × 5 cm, which was composed of 5 serially connected sub-cells with a 3.2 cm 2 aperture area for each. In a two-step patterning process with a femtosecond laser, the sub-cells were combined in a standard serial interconnection geometry, as shown in Fig. 5(a). After coating the module with hole transport material (HTM), a femtosecond laser was used to cut the ETL/perovskite/HTM layer, and then the electrode was evaporated to connect each sub-cell. Figure 5  Compared with a typical 1.2-cm 2 -area device, the module shows decreased J sc and FF values, which may be due to the interconnected width of the series resistance (a series resistance of 76 Ω) contributed by ITO. To verify the efficiency from the I-V curves, the steady-state output at a bias of 2.85 V was measured. The flexible module shows a nearly constant PCE of 7.6% during the measurement under the 100 mW cm −2 illumination shown in Supplementary Fig. S3.
A circular bending test was carried out on the flexible perovskite solar module to check the reliability. The module was tested with a radius of curvature of 32 mm in three different bending tests: bending parallel to the etching line, perpendicular to the etching line and both parallel and perpendicular to the etching line. The flexible devices exhibit decreased photovoltaic performance with bending the module, as shown in Fig. 5(c). The significant change in the flexible module performance upon multiple bending cycles originates from the decreased FF, which drops to 80% of its initial performance after 600 cycles in Supplementary Fig. S4. Most of the layers in the device, except for the ITO film, had high flexibility 34 . Specifically, the resistance of the ITO-PEN substrate was tested as a function of the number of bending cycles. The results clearly show a trend of increasing resistance with increasing number of bending cycles, as shown in Supplementary Table S2. This increasing resistance of ITO is in consistent with the decreased FF of the module.

Discussion
In conclusion, we have demonstrated a promising sequential deposition method for preparing CH 3 NH 3 PbI 3 films at an ultrafast PbI 2 evaporation rate of 20 Å s −1 , through which we obtained uniform and pinhole-free perovskite films in a short time. The rate-controlled PbI 2 deposition was investigated in detail, and devices were fabricated on flexible substrates. A 16 cm 2 flexible perovskite solar module based on the MAPbI 3 film deposited from an  optimized PbI 2 film was fabricated and had an efficiency of more than 8%. This strategy shows great potential for paving the way to the manufacture of large perovskite solar modules.
Preparation of devices. Indium tin oxide coated glass (ITO-Glass) and polyethylene naphthalate (ITO-PEN) were patterned by a femtosecond laser, followed by ultrasonic cleaning in detergent water, deionized water and ethanol for 15 min each. The substrates were then dried under an N 2 flow and treated with plasma for 5 min to remove any organic residues. Fullerene C 60 was thermally evaporated onto the ITO-coated substrates with a thickness of 10 nm under a deposition pressure of 8 × 10 -6 mbar at a deposition rate of 0.1 Å s −1 . The 160 nm PbI 2 films were then thermally evaporated on the C 60 underlayer with controlled deposition rates within the range between 0.5 Å s −1 and 40 Å s −1 , as monitored by a quartz crystal microbalance sensor. After the PbI 2 deposition process, the samples were transferred to a spin coater in an N 2 -filled glovebox for the subsequent chemical conversion steps. A 56 mg ml −1 methyl-ammonium iodide (MAI) solution in ethanol containing 24 μl of 2-methoxyethanol was spread on the whole PbI 2 film and then spin-coated at a rate of 3000 rpm for 30 s. The yellow film turned brown during the spin-coating and was then annealed at 150 °C on a hotplate outside of the glovebox for 10 min. To prepare the HTM layer, a solution of spiro-OMeTAD dissolved in ethyl acetate 35 (41.6 mg ml −1 , 17 μl ml −1 4-tert-butylpyridine, 7.8 μl ml −1 stock solution of 500 mg ml −1 lithium bis trifluoromethylsulfonyl imide in acetonitrile) was dropped onto the perovskite layer while spinning at a rotation rate of 2000 rpm. For the modules, P 1 patterning of the module was performed with a femtosecond laser (Wuhan Hongtuo). After the HTM layer was deposited, the C 60 /perovskite/spiro-OMeTAD layers were fully removed to ensure low contact resistance at the interconnected space through P 2 patterning. Finally, 10 nm MoO 3 and 60 nm Au were successively evaporated through a properly designed mask under high vacuum (4 × 10 −6 mbar) to complete the devices for tests.
Characterization. The photocurrent density-voltage characteristics of the devices were measured with a scan rate of 0.01 V s −1 under standard simulated AM 1.5 G illumination (100 mW cm −2 ) using a solar simulator (Oriel 94023 A, 300 W), which was calibrated using a standard Si solar cell (Oriel, VLSI standards). The external quantum efficiency (EQE) curves were obtained using monochromatic incident light produced by a power source (Pharos Technology) with a monochromator. Data acquisition was accomplished under DC mode with a power meter. The morphologies and microstructures of the deposited PbI 2 films and perovskite films were investigated using a field-emission scanning electron microscope (FE-SEM, Zeiss Ultra Plus) and an X-ray diffractometer (XRD, D8 Advance).